Lineage‐Specific Adaptive Strategies Shape the Response of de Novo Lager Yeast Hybrids to High‐Gravity Fermentation Authors: Sonia Albillos‐Arenal, Vasni Zavaleta, Fernanda Jeria, Darío González‐Venegas, Tomás A Peña, Pablo Villarreal, Francisco A Cubillos Published: 2026 ABSTRACT ABSTRACTLager beer is the most widely consumed fermented beverage worldwide and is produced by the yeast Saccharomyces pastorianus, which originated from hybridization between S. cerevisiae (Sc) and S. eubayanus (Se). Although the genetic factors underlying fermentation performance in de novo hybrids are well understood, how chimeric genomes evolve in relation to the parental background and stabilize under industrial stresses remains unclear. Here, we subjected representative novel hybrids derived from distinct parental lineages to long‐term experimental evolution for 250 generations under conditions mimicking high‐gravity brewing. We observed a clear pattern of the “rule of declining adaptability”, driven by diminishing‐returns epistasis, whereby hybrids with lower initial fitness exhibited the greatest evolutionary gains. Overall, 38 of 69 evolved lines produced more CO2 than their ancestral strains. Adaptive trajectories were strongly influenced by the Sc genetic background, while the Se contribution appeared comparatively limited, highlighting the dominant role of the Sc subgenome in determining brewing performance and evolvability. Genomic analysis of representative hybrids revealed different evolutionary strategies. The evolved HB3‐I line (beer Sc parent) showed a large‐scale chromosomal rearrangement with few coding‐sequence changes, consistent with fine‐tuning of an already adapted background. In contrast, HB44‐II (bioethanol Sc parent) accumulated multiple gene‐disruptive mutations affecting multiple pathways, indicating broader regulatory rewiring. These genomic changes were accompanied by transcriptional shifts in carbon metabolism, nutrient‐responsive pathways and membrane‐related functions. Together, our results provide a controlled framework for replaying yeast domestication, demonstrating that the adaptive potential of synthetic lager hybrids depends on their lineage, which is mainly shaped by the S. cerevisiae parental background. Introduction Lager beer is the most popular alcoholic drink worldwide, and its production depends on the fermentative activity of Saccharomyces pastorianus, an interspecific hybrid between Saccharomyces cerevisiae and the cold‐tolerant wild species Saccharomyces eubayanus (Baker et al. 2015; Hutzler et al. 2023). This yeast is specifically adapted to ferment cold wort containing sugars such as glucose, maltose and maltotriose under anaerobic, nutrient‐limited conditions typical of large‐scale beer production (Gibson et al. 2017). Yet despite its industrial success, S. pastorianus has limited genetic diversity, resulting from a single hybridization event and centuries of clonal propagation (Gallone et al. 2016; Gorter De Vries et al. 2019). This limited evolutionary history constrains its metabolic flexibility, stress tolerance and flavour diversity, traits that are currently sought after in the brewing industry (Gibson and Liti 2014). One such trait is high‐gravity brewing, which is being increasingly used to improve productivity, reduce water consumption and increase throughput (Puligundla et al. 2020). In this method, fermentation begins with concentrated wort (≥ 20°P), exposing yeast to intense osmotic stress, followed by ethanol levels exceeding 8% (v/v) that disrupt membrane function and redox balance (Puligundla et al. 2020; Thesseling et al. 2019). Under these conditions, many S. pastorianus strains exhibit slow fermentations, poor maltotriose attenuation and reduced viability, ultimately limiting their performance and sensory qualities in modern brewing (Brickwedde et al. 2017; Lei et al. 2012). Considering these limitations, generating new S. cerevisiae × S. eubayanus hybrids with high‐gravity tolerance offers a strong alternative for recovering and expanding the fermentative and adaptive traits that determine lager performance. De‐novo hybridization between S. cerevisiae and S. eubayanus strains has been used to create synthetic lager strains that leverage heterosis to outperform their parents in key brewing traits, including cryotolerance, aromatic complexity, sugar uptake and adaptive potential (Hebly et al. 2015; Krogerus et al. 2017; Zavaleta et al. 2024). However, as recently demonstrated by Zavaleta et al. (2024), hybridization alone rarely translates into improved fermentation kinetics under industrial lager conditions. This limitation indicates that interspecific crosses require subsequent evolutionary fine‐tuning to overcome regulatory bottlenecks and achieve the robustness needed for high‐gravity brewing (Martínez and Lang 2026; Molinet et al. 2024; Zavaleta et al. 2024). Experimental evolution is a powerful strategy to unlock latent adaptive potential in industrial yeasts (Mardones et al. 2022; Molinet et al. 2024; Pinto and Stelkens 2025). By exposing populations to prolonged selective pressures, this approach reveals beneficial mutations that drive adaptive regulatory shifts, ultimately enhancing performance traits such as ethanol tolerance, osmoadaptation and carbon metabolism (Lang et al. 2013; Mardones et al. 2022; Molinet et al. 2024; Pinto and Stelkens 2025). In interspecific hybrids, genomic plasticity, including frequent structural variation, loss of heterozygosity (LOH) and aneuploidy, often accelerates evolution (D'Angiolo et al. 2020; Smukowski Heil et al. 2019). However, this chimeric architecture can also constrain the rate of adaptation if beneficial phenotypes depend on complex structural events, such as specific LOH tracts, that occur at limiting rates (Bautista et al. 2024). These mechanisms mirror those found in natural yeast domestication, which Sampaio and Pontes (2025) recently redefined as a multi‐stage process involving wild, industrial transitions shaped by human‐imposed selective bottlenecks (Sampaio and Pontes 2025). Their framework highlights how repeated cycles of selection in anthropogenic environments have reshaped yeast genomes, often via regulatory streamlining and stress specialisation. However, the hybrid's chimeric genomes, assembled from two distinct regulatory architectures, are susceptible to cis/trans and protein incompatibilities, in which promoters and transcription factors from different species fail to interact effectively (Bendixsen et al. 2022; Martínez and Lang 2026). This regulatory configuration could affect the coordination of global gene expression programs required for robust high‐gravity fermentation. Building on this perspective, we subjected representative S. cerevisiae × S. eubayanus hybrids to experimental evolution under high‐gravity fermentation. This method allowed us to replicate the combined osmotic, ethanol and nutrient stresses typical of lager industrial brewing environments. Comparative phenotypic and genomic analyses showed that adaptive trajectories depended on hybrid lineage and ploidy, resulting in reproducible but distinct evolutionary outcomes. Overall, our study demonstrates that laboratory evolution can mimic key aspects of yeast domestication and provides insights into how environmental stress impacts hybrid genomes, offering a foundation for intentionally improving lager strains. Materials and Methods Yeast Strains and Culture Media A total of 21 hybrid strains resulting from crosses between S. cerevisiae and S. eubayanus, previously generated in Zavaleta et al. (2024), were studied, alongside their parental lines, comprising 5 S. cerevisiae and 6 S. eubayanus strains. The S. cerevisiae strains originated from various anthropogenic environments, reflecting diverse ecological backgrounds (Peter et al. 2018). The S. eubayanus strains were sourced from different sites in Chilean Patagonia and were selected for their robust fermentative performance and for representing the main Patagonia‐B genetic lineages (PB‐1, PB‐2, PB‐3) and the Admixed lineage (Nespolo et al. 2020). For comparative purposes in fermentation assays, the commercial lager yeast S. pastorianus Saflager W34/70 (Fermentis, France) was used as a reference strain. All yeast strains were propagated on YPD agar (1% yeast extract, 2% peptone, 2% glucose and 2% agar) and preserved at −80°C in 20% glycerol stocks. The complete list of strains utilised in this study is provided in Tables S1 and S2. Beer Wort Fermentations and Sugar Consumption Estimation Small‐scale fermentations were performed using high‐gravity wort at 20°P. The wort was pre‐oxygenated to 15 mg/L and supplemented with 0.3 ppm ZnCl2, (Mardones et al. 2020) and all experiments were conducted in triplicate. Prior to fermentation, yeast cultures were pre‐grown in 5 mL of pasteurised (100°C, 40 min) 12°P malt extract (Maltexco, Chile) at 20°C for 24 h. Subsequently, cells were acclimatised to high‐gravity conditions by transferring them to 45 mL of 20°P wort for an additional 24 h. Fermentations were inoculated at a starting cell concentration of 3 × 107 cells/mL (equivalent to a total of 1.5 × 109 cells into 50 mL of fresh wort). This pitching rate was calculated based on the wort density and volume, following standard high‐gravity brewing recommendations (White and Zainasheff 2010). Fermentations were conducted at 12°C for 14 days without agitation. Cultures were performed in bottles with loosely fitted caps to allow continuous CO2 release while minimising pressure buildup. We used an identical vessel configuration across all conditions, together with negative controls lacking yeast, to reliably estimate that weight loss reflected relative CO2 production. Daily weight‐loss measurements were used to monitor CO2 evolution as an indicator of fermentation progress. Fermentation performance was evaluated by calculating total CO2 production and the area under the curve (AUC) using the R package Growthcurver. Residual sugar concentrations were measured at Days 1, 3, 6, 9 and 14 of micro‐scale fermentations. For each time point, 1.5 mL of fermented must was centrifuged, and the supernatant was recovered. Residual sugars were quantified by HPLC (Shimadzu Prominence system; Shimadzu, USA) equipped with a refractive index detector (Mardones et al. 2022). Filtered samples (0.22 μm syringe filters) were injected (20 μL) and separated on an Aminex HPX87H column using 5 mM H2SO₄ with 4 mL/L acetonitrile as the mobile phase (flow rate: 0.5 mL/min). Experimental Evolution of Hybrid Strains Hybrid yeast strains underwent experimental evolution in high‐gravity wort (20°P), which was prepared from malt extract, and its density was verified using a refractometer. The medium was further supplemented with 6% (v/v) ethanol to impose additional selective pressure. For each experimental line, 1 mL of this medium was distributed into individual wells of 24‐well microplates. Each hybrid strain was inoculated in triplicate, resulting in a total of 27 independent evolutionary lineages. Cultures were serially transferred to fresh medium every 72 h, maintaining an inoculation density of 1 × 106 cells per well, and this process was continued for a total of 250 generations. The number of generations occurring between each transfer was calculated according to the formula described by (Mardones et al. 2022): Upon completion of 250 generations, the evolved populations were cryopreserved in 20% (v/v) glycerol at −80°C for long‐term storage. Each evolved lineage was labelled using the ancestral strain's identifier, followed by a Roman numeral (I, II, III) to denote biological replicates (e.g., HB2 I, HB2 II, HB2 III for the three replicates derived from the HB2 ancestor; see Table S2). Phenotypic Profiling of the Evolved Hybrid Strains To evaluate the adaptive fitness of the evolved hybrid strains, growth kinetics were examined both in the experimental evolution medium and under related stress conditions. Initially, strains were pre‐cultured in 200 μL of 0.67% yeast nitrogen base (YNB) medium supplemented with 2% glucose in 96‐well plates at 20°C for 48 h. Following this, 5 μL aliquots of each pre‐culture were inoculated in triplicate into 0.67% YNB containing: (i) 2% maltose and 6% ethanol, (ii) 20% maltose, or (iii) 20% maltose plus 6% ethanol. Growth was continuously monitored by measuring optical density at 620 nm (OD₆2₀) every 30 min until cultures reached the stationary phase, using a TECAN Sunrise microplate reader (Mardones et al. 2022). The maximum growth rate (μ_max) was derived from the slope of the exponential growth phase using the Growthcurver package in R. To assess relative fitness improvements, μ_max values of evolved strains were normalised against their corresponding ancestral strains according to: μMax relative = ((μMax evolved—μMax ancestral))/(μMax ancestral). Statistical significance of differences in μMax was determined using Student's t‐tests implemented in R, with a threshold of p‐value < 0.05 indicating significant changes. Illumina Read Processing and Consensus Variant Calling Genomic DNA was extracted using Zymo Quick‐DNA bacterial/fungal extraction kit (cat. no. D6007, Zymo Research) following the manufacturer's instructions. The library prep reaction was a 100× miniaturised version of the Illumina Nextera method. Samples were sequenced on a NextSeq 500/550 High Output Kit v2.5 (300 Cycles) flow cell following the manufacturer's instructions. Raw reads quality was checked using FastQC (Andrews 2010). Paired‐end Illumina reads were processed with fastp v0.23.4 (−q 30) (Chen et al. 2018). For each evolved line, three independent single‐colony isolates (clones) were sequenced. Trimmed paired‐end Illumina reads of evolved hybrids were aligned independently to the corresponding ancestral genome assemblies using BWA‐mem (Li 2013). Mapping quality and summary statistics were obtained using Qualimap v2.3 (García‐Alcalde et al. 2012). SAM files were sorted and transformed into BAM format using Samtools 1.21 (Danecek et al. 2021). Read‐group tags (ID, SM, LB, PL) were then added using GATK v4.3.0.0 (AddOrReplaceReadGroups), duplicates were identified and flagged with GATK v4.3.0.0 (MarkDuplicates) (DePristo et al. 2011), and the resulting BAMs were re‐indexed with Samtools 1.21. Variant calling was done using GATK v4.3.0.0 HaplotypeCaller (default settings), calling variants per chromosome. VCF files were filtered by removing missing data (−max‐missing 1) and minimum site quality of 30 (−minQ 30) in in vcftools 0.1.17 (Danecek et al. 2011). To address clonal heterogeneity within each evolved lineage and to obtain high‐confidence mutations, consensus variants were retained from each clone only if FORMAT/DP ≥ 20 (Danecek et al. 2021). Importantly, to account for subgenomic ploidy differences, variants located within the S. cerevisiae subgenome were additionally required to have no reference‐supporting reads (i.e., FORMAT/AD reference count = 0); this strict restriction was not applied to the S. eubayanus subgenome in HB44. Besides, an alternative filtered dataset was generated comprising clone‐level SNPs. Variants were retained if they had a read depth (DP) ≥ 10 in at least one clone carrying the variant. We removed any 500‐bp windows with depth≥max (40×, 1.5× mean depth), and calculated the union across the three clones. Dense variant clusters were excluded by removing all variants falling within intervals containing ≥ 3 variants within a 300‐bp window. To assign variants to the parental subgenomes and reduce the prevalence of false positives arising from non‐specific mapping between highly similar regions, reads from each clone were also mapped separately to each ancestral subgenome assembly ( S. cerevisiae and S. eubayanus), followed by variant calling and filtering as above. Subgenome‐specific consensus VCFs (per evolved line) were generated by intersecting the three clones with BCFtools isec, then intersecting the result with the consensus call set from the whole‐genome ancestral assembly, retaining only concordant sites. The resulting ultra‐filtered VCFs were annotated and used for effect prediction with SnpEff (Cingolani et al. 2012), which classifies variant impacts (e.g., high/moderate/low/modifier) based on the affected genomic feature and predicted functional consequence. Nanopore Sequencing, Hybrid Genome Assembly and Structural Variant Calling High molecular weight DNA was extracted using a Qiagen Genomic‐tip 20/G kit (Qiagen, Hilden, Germany) as previously described by (Villarreal et al. 2024). Following the manufacturer's protocol, libraries were sequenced on a MinION R10.4 flow cell (Oxford Nanopore Technologies, UK). The raw pod5 files were transformed to fastq files and debarcoded using Dorado v0.7.2 with the “super high accuracy” model (https://nanoporetech.com/accuracy). Barcodes and adapter sequences were trimmed using Porechop (https://github.com/rrwick/Porechop) and filtered with Filtlong (https://github.com/rrwick/Filtlong) using a Phred score of 15. In addition, we used publicly available paired‐end Illumina sequence data previously generated for each strain. Flye v2.9 (Kolmogorov et al. 2019) was used for assembly, followed by a round of medaka v2.0.1 (medaka: Sequence correction provided by ONT Research, 2019) and pilon v1.24 (Walker et al. 2014) for long and short read polishing. Polished assemblies' contigs were reconstructed with RagTag v2.1.0 (Alonge et al. 2022) using as reference a concatenated reference genome of S. eubayanus CBS12357T (Libkind et al. 2011) and S. cerevisiae S288c. Contig integrity was assessed using D‐GENIES (Cabanettes and Klopp 2018). Hybrid genome assemblies were annotated with Liftoff v1.6.3 (Shumate and Salzberg 2024) using S. eubayanus strain FM1318 (GCA_001298625.1) as a reference for the S. eubayanus subgenome, and S. cerevisiae strain S288c (GCA_000146045.2) for the S. cerevisiae subgenome. The annotation files were transformed from.gff to .gbk format and used to build annotation databases in SnpEff (Cingolani et al. 2012). Evolved hybrid assemblies were compared with ancestral hybrid assemblies using MUMmer4 (Marçais et al. 2018) and structural variants (SVs) were called using MUM&Co v3.8 (O'Donnell and Fischer 2020). Transcriptome Analysis To investigate the gene expression patterns underlying the fermentative performance of hybrids in high‐gravity wort, we performed RNA sequencing (RNA‐seq) analysis. Samples were collected after 24 h of fermentation from de novo hybrids exhibiting distinct sugar consumption profiles: HB3‐A, HB3‐I, HB44‐A and HB44‐II. RNA was obtained from triplicate cultures using the E.Z.N.A Total RNA kit I (OMEGA) as previously described (Molinet et al. 2024; Venegas et al. 2023). Total RNA was recovered using the RNA Clean and Concentrator Kit (Zymo Research). The cDNA libraries were generated using the TruSeq RNA Sample Prep Kit v2 (Illumina, San Diego, CA, USA). Illumina libraries from hybrids were generated using paired‐end 150 bp reads on a NextSeq 500/550 High Output Kit v2.5 (300 Cycles) flow cell following the manufacturer's instructions (Venegas et al. 2023). Reads were quality‐trimmed and filtered using Trimmomatic (version 0.39) with the following parameters: ILLUMINACLIP (2:30:10) to remove adapter sequences, LEADING:20 and TRAILING:20 to trim low‐quality bases from the ends, SLIDINGWINDOW:4:20 for dynamic trimming based on average quality, and MINLEN:36 to discard short reads. Trimmed reads were mapped to the corresponding reference genomes HB3 and HB44 using HISAT2 (version 2.2.1) (Kim et al. 2019). Paired‐end trimmed reads were mapped using default settings and multithreaded execution. SAM output was converted to BAM and sorted using SAMtools (version 1.15). Finally, the resulting BAM files were sorted and indexed using the SAMtools (Li et al. 2009). Differentially expressed gene (DEG) analysis was performed in R using the DESeq2 package v4.1.2 (Love et al. 2014). DEGs were thus identified in all conditions, and the ones with (p‐adj < 0.1 and log2FC > 1; log2FC < −1) were subsequently extracted for further analysis. Metabolic pathway enrichment was analysed using the enrichKEGG and enrichGOtools (ClusterProfiler (Wu et al. 2021)) based on gene identifiers in Entrez format. Analyses were performed separately for each hybrid ancestral strain and each evolved strain. Transcription factors associated with the statistically differentially expressed genes were analysed using YETFASCO (De Boer and Hughes 2012), with all DEGs used as the background. Subsequently, YEASTRACT (Teixeira et al. 2023) was used to assign transcription factors to their target genes. Results S. cerevisiae Parental Background Predominantly Determines Hybrid Performance Under High‐Gravity Fermentation To assess hybrid performance under high‐gravity brewing stress, we selected 21 representative S. cerevisiae × S. eubayanus hybrids previously characterised at lower gravity conditions by Zavaleta et al. (2024), which were generated by combining different lineages from both species. The hybrids, along with their parental strains (Tables S1 and S2), were evaluated under high‐gravity fermentation conditions (20°P, 12°C). Fermentation progress was monitored using cumulative CO2 loss as a proxy for metabolic activity, with the commercial lager strain W34/70 included as an industrial control (Table S3a). Significant interspecific differences were observed, with S. eubayanus strains displaying lower fermentative capacity, while S. cerevisiae strains and the hybrids reached performance levels comparable to the W34/70 reference strain (p‐value > 0.05, one‐way ANOVA with Tukey's HSD post hoc test; Table S3b, Figure 1A). Nevertheless, notable intraspecific variation was observed within the S. cerevisiae isolates and the new hybrids. Among the S. cerevisiae parents, the beer‐related strain Sc_Beer_2 (57.5 ± 0.4 g/L CO2 loss) performed similarly to W34/70 (56.5 ± 0.4 g/L CO2 loss); whereas bioethanol, wine and sake‐derived Sc_BEth_2 (45.7 ± 0.5 g/L), Sc_Wine_2 (45.9 ± 1.6 g/L), and Sc_Sake_1 (21.7 ± 1.5 g/L) showed significantly lower CO2 production than W34/70, respectively (p‐value < 0.05, one‐way ANOVA with Tukey's HSD post hoc test; Table S3c). Hybrid fermentation phenotypes largely reflected the performance of their S. cerevisiae parent strain (Figure 1B; Table S4). Hybrids derived from the high‐performing beer strain Sc_Beer_2 tended to exhibit the strongest fermentative capacity, whereas hybrids derived from the bioethanol, wine and sake lineages showed lower fermentative performance (Table S3c). Only the hybrids derived from Sc_Sake_1 showed a significant increase in fermentative performance relative to their parent, likely reflecting the exceptionally low performance of the parental strain. Nevertheless, none of the hybrids surpassed the commercial lager strain W34/70. In contrast, the S. eubayanus parental contribution appeared functionally recessive, and nearly all hybrids outperformed the complete S. eubayanus parental range (Figure 1C). Fermentative capacity of the ancestral hybrids and their parental S. cerevisiae and S. eubayanus strains. (A) CO2 production levels in S. cerevisiae (pink), S. eubayanus (yellow), hybrids (grey) and the lager commercial W34/70 (blue), under 20°P wort. The names of the selected strains for the experimental evolution assay are indicated. (B) CO2 production of Hybrids depending on the S. cerevisiae parent. Hybrids are coloured according to their S. cerevisiae lineage of origin: Beer (green), Sake (blue), Bioethanol (orange) and Wine (red), with W34/70 (blue) as a control. (C) Similarly, each Hybrid, depending on the S. eubayanus parent, is plotted. Hybrids are grouped by S. eubayanus lineage: PB‐1 (purple), PB‐2 (light blue), PB‐3 (light green) and Admixed (gold). Blue asterisks indicate significant differences compared with the commercial control W34/70. Black asterisks denote statistical differences of fermentative performance between each hybrid group and its corresponding parental strains, performed using analysis of variance (ANOVA) followed by Tukey's HSD test. Together, these findings suggest that fermentative vigour in S. cerevisiae × S. eubayanus hybrids primarily derives from the S. cerevisiae genetic background, highlighting its dominant role in shaping brewing performance. Notably, only hybrids derived from Sc_Sake_1 (a lineage whose fermentative capacity closely resembles that of S. eubayanus) outperformed their S. cerevisiae parent in high gravity wort (Table S4). In contrast, hybrids from the Sc_BEth cluster showed the lowest fermentative performance, suggesting that hybridization alone is insufficient to enhance fermentation under high‐gravity wort conditions, while highlighting their potential for improvement. Therefore, we selected six high‐performing hybrids (derived from the S. cerevisiae Sc_Beer_2) and two low‐performing hybrids (derived from Sc_BEth_2) (highlighted by name in Figure 1A) for long‐term experimental evolution under high‐gravity and ethanol stress. Experimental Evolution Under Brewing Stress Drives Rapid Adaptation in Yeast Hybrids, With the Largest Gains in Initially Poor Performers To improve the high‐gravity wort fermentation phenotype in selected hybrids, we conducted a long‐term evolution experiment. Each hybrid was evolved in triplicate, yielding three lines (I, II and III), except for the HB44 line III, which was excluded from further analyses due to contamination (Figure S1). Following 250 generations of experimental evolution in 20°P wort supplemented with 6% ethanol, we assessed the adaptive changes in all lines of the evolved hybrids by measuring their maximum growth rates (μmax) under three stress scenarios: 2% maltose combined with 6% ethanol (ethanol toxicity with moderate osmotic stress, hereafter MALEtOH), 20% maltose (high osmotic stress, hereafter HiMAL), and 20% maltose with 6% ethanol (combined osmotic and ethanol stress, hereafter HiMALEtOH) (Table S5). Regarding HiMAL, 7 of the 23 evolved lines significantly outperformed their ancestors, while 6 showed lower performance (Figure 2A). Interestingly, the evolved lines derived from the hybrids HB37 and HB44, which originate from the S. cerevisiae bioethanol lineage, showed the most significant improvements after experimental evolution (Figure 2B), although these hybrids initially performed poorly in high‐gravity fermentation (Table S6a). In contrast, evolved hybrid lines from the beer lineage remained the top performers (Table S6b), displaying only modest gains, consistent with their already superior growth rates compared to the commercial strain W34/70. (Figure S2). When the high maltose and ethanol stress were combined (HiMALEtOH), 9 evolved lines demonstrated significant higher adaptive improvements, and 2 lower (Figure 2A; Table S6a). Notably, both the ancestral and evolved hybrids derived from the S. cerevisiae beer lineage outperformed the commercial control W34/70 in growth performance with HiMALEtOH stress (Figure S2). In the MALEtOH condition, 12 evolved lines showed a significant increase in μmax compared to their ancestors, and 4 showed a decrease (Figure 2A). Evolved lines from hybrids HB15‐III and HB37‐I showed the most pronounced improvements, with Δμmax of 0.05 h−1, in both cases (Figure 2A). Evolutionary gains were more pronounced in hybrids derived from S. cerevisiae bioethanol lineages than in those derived from beer lineages (Figure 2B). Specifically, experimental evolution enabled several lines from bioethanol‐lineage hybrids to exceed the μmax of the commercial strain W34/70, overcoming the growth deficit observed in their ancestral states (Figure S2). Hybrid's phenotypic profiles after long‐term evolution. (A) Heatmap showing changes (Δ) in maximum growth rate (μmax) of evolved hybrids relative to their ancestral strains in high‐gravity wort–mimicking media: HiMAL (20% maltose), HiMALEtOH (20% maltose +6% ethanol) and MALEtOH (2% maltose +6% ethanol). Asterisks indicate significant differences (p < 0.05, Student's t‐test). (B) μmax values of the evolved hybrids, depending on the S. cerevisiae lineages (beer or ethanol), were normalised to the ancestral ones in the three different stress conditions, with the ancestral value set to 0 as the reference. Overall, hybrids HB12, HB3, HB37 and HB44 exhibited the most consistent improvements across stress conditions, demonstrating that experimental evolution, particularly under MALEtOH, can rapidly enhance the fitness of initially poor S. cerevisiae × S. eubayanus hybrid performers in brewing environments. Increased Fermentation Performance Depends on the S. cerevisiae Lineage, With Constrained Evolvability in Beer‐Derived Hybrids To evaluate whether experimental evolution enhanced high‐gravity fermentation, we compared the fermentative performance of the evolved hybrid lines with their respective ancestors, using the industrial lager strain W34/70 as a benchmark (Table S7a). Across all hybrids, 5 out of 23 evolved lines showed increased CO2 production relative to their progenitors, with the significant improvements observed in HB3‐I, HB44 (I and II), and HB37 (I and III) (Figure 3A; Table S7b). Remarkably, HB44‐II (derived from the bioethanol S. cerevisiae parent) displayed the greatest relative enhancement, achieving a 24% increase in CO2 release. In contrast, beer‐derived hybrids exhibited only modest gains, ranging from 1% to 10% in HB3 and HB2, and even showed reductions of up to 10% in HB6, HB9, HB12 and HB15. Then, we asked whether the parental lineage influenced the magnitude of improvement following evolution. Analysis of relative CO2 loss revealed that increases in fermentative performance strongly depended on the S. cerevisiae background: hybrids carrying bioethanol‐derived S. cerevisiae parents exhibited significantly greater improvements than those derived from beer lineages (Figure 3B; Table S7c). By contrast, hybrids harbouring S. eubayanus parents from different lineages (PB‐1, PB‐2, PB‐3 or admixed) did not differ significantly from each other (Figure 3C). Together, these results support the “constrained evolvability” hypothesis: beer yeasts, already specialised for brewing, likely had fewer adaptive pathways available. Fermentation capacity assessment of evolved S. cerevisiae × S. eubayanus hybrids. (A) Microfermentations were performed in beer wort at 20°C for 14 days, with measurement of CO2 loss. Relative CO2 loss indicates the difference in total CO2 released (after 14 days of fermentation) between each evolved hybrid line and its ancestral strain. A lineage‐based analysis was also conducted to compare performance differences between hybrids with (B) S. cerevisiae and (C) S. eubayanus. (D) Enhanced sugar utilisation was monitored during high‐gravity fermentation by evolved hybrids of S. cerevisiae × S. eubayanus. Statistically significant differences are marked by asterisks (*) (p < 0.05, Student's t‐test). Variations in fermentation performance are closely linked to differences in sugar consumption. To examine how evolutionary adaptation affects sugar utilisation in yeast during beer wort fermentation, three representative evolved lines were selected: HB3‐I, derived from a beer S. cerevisiae parent, and HB37‐I and HB44‐II, both derived from a bioethanol S. cerevisiae strain. These hybrid lines were selected because they exhibited the greatest increases in fermentative capacity relative to their respective ancestral strains and the prior phenotypic characterisation. Each hybrid line was fermented again, this time using samples collected throughout fermentation to measure residual glucose, maltose and maltotriose (Table S8a). Glucose was fully consumed by all strains within 70 h; therefore, comparative analyses were restricted to maltose and maltotriose, where the most pronounced phenotypic differences were observed. The evolved HB3‐I showed significantly improved maltose and maltotriose consumption compared to the ancestral hybrid, with overall increases of 43% and 175%, respectively, by the end of fermentation (Figure 3D, Table S8b). The evolved HB37‐I line showed a modest yet significant increase in maltose consumption after 143 h, with no significant difference observed by the end of fermentation (Table S8c). In contrast, the evolved HB44‐II hybrid showed a 281.16% increase in maltose consumption at the end of fermentation (Figure 3D, Table S8b). These results demonstrate that evolutionary adaptation can significantly enhance the utilisation of key wort sugars. Accordingly, hybrids HB44‐II and HB3‐I were selected for further genomic and transcriptomic analyses to identify the molecular basis of these improvements. Adaptive Evolution Exploits Distinct Structural and Mutational Mechanisms in Hybrid Yeast Genomes To verify parental origin, assess ploidy level and detect nucleotide‐level variation arising during experimental evolution, the evolved strains HB3‐I and HB44‐II and their respective ancestors (HB3‐A and HB44‐A) were sequenced using Illumina short‐read technology. Ploidy analysis confirmed that the evolved hybrids retained the same Se:Sc subgenome ratios as their ancestral counterparts, with a 1:1 configuration in the diploid HB3‐I and a 2:1 configuration in the triploid HB44‐II (Figure S3). While experimental evolution in yeast can often lead to rapid changes in ploidy or subgenome composition, our results indicate that after 250 generations, the overall genomic architecture remained stable under the specific conditions and duration of this study. To characterise the mutational landscape of the evolved hybrids, we then examined the accumulation of single‐nucleotide polymorphisms (SNPs) across both subgenomes. We identified 20 and 62 SNPs in the HB3‐I and HB44‐II evolved lines, respectively. Our analysis revealed a striking divergence in the impact of the mutational profiles between the two lineages, HB3‐I exhibited a highly minimalist landscape (Table S9a). When applying more relaxed filter, it harboured a handful of upstream and downstream modifier variants (Figure 4A, Table S9a). Within the S. cerevisiae subgenome, these modifiers were associated with genes such as the mitochondrial assembly factor COI1; the vacuolar protein sorting VPS13; the nucleolar protein essential for the early steps of 35S rRNA processing, FCF2; and APM4, which facilitates the uptake of membrane proteins. S. eubayanus subgenome was similarly minimalist, harbouring only modifier‐class mutations and no coding changes across all three clones. In contrast, HB44‐II accumulated a richer, more functionally consequential set of mutations in both genomic backgrounds (Figure 4B, Table S9a). The S. cerevisiae subgenome carried a high‐impact frameshift in SRD1 (rRNA processing), together with moderate‐impact missense variants in the RNA‐processing genes SPP41 and PRP21, and in DPH2 and NMT1, which influence protein modification and lipid acylation. The S. eubayanus subgenome also contained a high‐impact frameshift in PCT1, a key enzyme in phosphatidylcholine biosynthesis (involved in membrane metabolism), as well as additional modifier variants. Genomic analysis of evolved hybrids. (A) Summary of genes affected by SNPs identified from Illumina short‐read data and annotated for predicted functional impact using SnpEff (black), as well as key genes located in regions containing structural variants (SVs) detected by whole‐genome alignments with nucmer and identified using MUM&Co v3.8, in the evolved hybrid HB3‐I relative to HB3‐A (B), and in the evolved hybrid HB44‐I relative to HB44‐A. Additionally, Oxford Nanopore long‐read sequencing of the ancestral hybrids and the most improved evolved line from each background was performed to enable de novo genome assembly and detection of structural variants (SVs). The high quality and completeness of these assemblies were confirmed through BUSCO analysis, which detected over 99.6% of the expected orthologs in all cases (Table S10a). In this way, we sought to identify SVs arising during experimental evolution, classify them by variant type, and examine the genes located within the affected regions. In total, HB3‐I carried 18 SVs, whereas HB44‐II accumulated 22 events relative to the ancestral hybrid (Table S10b; Figure S4). The most prominent genomic change in HB3‐I was a ~70 kb inversion on chromosome III of Se, which affected the orientation of more than 40 genes without disrupting any coding sequences (Figure 4A). These included the silent mating‐type loci HMLα1 and HMLα2, possibly reflecting residual genomic flexibility linked to mating‐type switching during hybridisation. In the Sc subgenome, HB3‐I carried two small ORF‐breaking insertions affecting MIL1 (YFL034W) and MMP1 (YLL061W) (Figure 4A). Regarding HB44‐II, in the Sc subgenome, several SVs resulted in the complete or partial loss of coding sequences (Figure 4B). These included a deletion of the PIR1 (YKL164C) coding region, which encodes a cell wall–associated structural protein. HB44‐II Sc also carried a structural duplication encompassing GAL80 (YML051W) and AIM32 (YML050W). Within the Se subgenome, HB44‐II exhibited a complete gene‐loss event, resulting in the full deletion of TDA10, which participates in carbohydrate, amino acid and lipid metabolism. A second deletion partially disrupted the PIR1 ortholog, providing convergent loss‐of‐function across subgenomes for this gene family. In this way, HB3‐I and HB44‐II followed markedly different evolutionary trajectories: HB3‐I underwent a large‐scale rearrangement and several minor coding‐sequence disruptions, whereas HB44‐II accumulated multiple gene‐disruptive mutations targeting pathways related to nutrient sensing, cell wall integrity and genomic stability. Transcriptomic Profiling Reveals Genotype‐Specific Gene Expression Changes During High‐Gravity Wort Fermentation To determine whether the phenotypic improvements observed after experimental evolution were accompanied by changes in gene expression, we performed RNA sequencing (RNA‐seq) on selected ancestral and evolved hybrids (HB3‐A, HB3‐I, HB44‐A and HB44‐II). Transcriptomes were profiled after 24 h of high‐gravity fermentation. In the HB3 background, 6 genes were significantly upregulated in the evolved line (HB3‐I versus HB3‐A), whereas 2 genes were significantly downregulated (Figure 5A; Table S11a; p‐adj < 0.1 and log2FC > 1; log2FC < −1). Interestingly, several genes involved in carbohydrate metabolism and transport were upregulated in the evolved HB3‐I line, including HXT4 and AQY3. The aromatic amino acid biosynthesis gene ARO3, which catalyses the first step of this pathway, also showed increased expression compared with the HB3‐A hybrid. Furthermore, a GSEA of the complete, fold‐change‐ranked gene list demonstrated a significant positive enrichment for carbohydrate transmembrane transporter activity (Table S11b). In contrast, genes downregulated in HB3‐I were primarily associated with cell division, mitotic cell cycle and reproductive processes. Differential gene expression and gene regulatory networks in HB3‐I and HB44‐II compared with their ancestral counterparts. Volcano plot of RNA‐seq results for (A) HB3‐I versus HB3 (ancestral) and (B) HB44‐I versus HB44 (ancestral). Top‐ranked Transcription factors (TFs) with significant associations to DEGs in (C) HB3‐I versus HB3‐A and (D) HB44‐I versus HB44‐A. HB44‐II showed a more balanced transcriptional shift, with 10 genes upregulated and 10 downregulated in the evolved line (HB44‐II vs. HB44‐A) (Figure 5B; Table S12a). We identified upregulated genes associated with carbon catabolite regulation of transcription, such as GAL1 and MIG3, which are linked to nutrient response pathways. Other notable upregulated genes in the evolved line included FUR4 (plasma membrane uracil permease), UIP3 (putative integral membrane protein) and the phosphatase POA1. Downregulated genes were significantly associated with biological processes such as DNA damage response, cellular response to chemical stimuli, DNA repair and chromosome organisation (Figure 5B). Crucially, when performing a GSEA we observed a broad enrichment of ‘DNA‐binding’ functions, indicative of a widespread regulatory fine‐tuning (Table S12b). These results indicate that adaptive improvements in HB44‐II are primarily associated with regulatory rewiring of carbon metabolism and nutrient‐responsive pathways, accompanied by reduced investment in genome maintenance processes. Finally, top‐ranking transcription factors were identified based on YETFASCO scores and the number of differentially expressed genes (DEGs) they regulate (Figure 5C,D). Yap1p was the most prominent transcription factor in both HB3‐I and HB44‐II, uniquely regulating all significant DEGs and achieving the highest score in both analyses. In addition, Sko1p was common in both hybrids. These results indicate that adaptive evolution promotes genotype‐specific transcriptional changes that converge functionally on enhanced carbon metabolism and nutrient utilisation during high‐gravity fermentation. Discussion High‐gravity fermentation increases ethanol yield and reduces processing costs; however, it imposes severe osmotic and ethanol stresses that require robust yeast physiology (Puligundla et al. 2020; Gibson et al. 2017). Therefore, to enhance industrial fitness, we evolved 23 hybrids for 250 generations in high‐gravity wort and ethanol stress. This process led to enhanced fermentation kinetics, improved stress tolerance and more efficient sugar metabolism, particularly in hybrids derived from bioethanol S. cerevisiae backgrounds. These hybrids, which initially exhibited poor fermentative performance, showed the greatest gains during experimental evolution, suggesting that lower baseline fitness may offer greater adaptive potential. This observation aligns with previous studies showing that strains or populations starting from lower initial fitness tend to exhibit more pronounced improvements during adaptation (Vega‐Macaya et al. 2025). This phenomenon is often called the ‘rule of declining adaptability’, where in a specific environment, genotypes with lower fitness are repeatedly observed to adapt more quickly than their fitter counterparts (Couce and Tenaillon 2015). Mechanistically, this pattern is driven by diminishing‐returns epistasis, a global constraint where the effect size of new beneficial mutations systematically decreases in fitter genomic backgrounds along an evolutionary trajectory (Chou et al. 2011; Wiser et al. 2013; Wünsche et al. 2017). Remarkably, recent evidence demonstrates that this global‐epistasis scaling remains robust and predictable across diverse environments (Ardell et al. 2024), thereby fundamentally constraining the rate of further adaptation in our highly fit ancestral lineages. Lager yeast (S. pastorianus), a natural hybrid of S. cerevisiae and S. eubayanus, exemplifies the potential of interspecies hybridization to yield strains with desirable industrial traits. While S. cerevisiae × S. eubayanus hybrids show heterosis and improved traits (Krogerus et al. 2016; Steensels et al. 2014), our results indicate that they did not reach optimal performance under high‐gravity conditions. We built upon the work of Zavaleta et al. (2024), who generated and characterised de novo S. cerevisiae × S. eubayanus hybrids. Their evaluation focused on standard wort fermentations, in which hybrids showed high phenotypic diversity and notable heterosis in aroma compound production, particularly in beer S. cerevisiae ‐derived hybrids, but no significant heterosis in fermentation performance. In the present study, we tested these same hybrids under more demanding high‐gravity conditions, and we still observed that their fermentative performance largely reflected the traits of their S. cerevisiae parent strains, with S. eubayanus contributions being functionally recessive. Hybrids derived from high‐performing S. cerevisiae strains matched industrial benchmarks, while those from low‐performing parents showed reduced fermentation capacity, indicating that glycolytic efficiency is mainly governed by the S. cerevisiae subgenome. Interestingly, the dominance of S. cerevisiae was also observed during the experimental evolution assay, where bioethanol‐derived hybrids exhibited greater potential for fitness increases (evolvability). The specifics of sugar utilisation differed between lineages: beer‐derived hybrid HB3‐I displayed faster CO2 release and enhanced maltose and maltotriose consumption during late fermentation, whereas bioethanol‐derived hybrid HB44‐II exhibited improved maltose and maltotriose utilisation during mid and late fermentation stages. Genomic and transcriptomic analyses support these observations: In HB3‐I, despite a low number of high‐stringency genomic variants, specific SNPs and SVs led to the upregulation of carbohydrate transporters (HXT4/AQY3/HXT2/HXT10/STL1), which likely facilitated early‐stage sugar uptake. In contrast, in the HB44‐II background, the evolved strain displayed a substantially greater number of SNPs and SVs, once again highlighting the previously discussed phenomenon of diminishing‐returns epistasis. The duplication of GAL80 probably influenced the differences observed by transcriptional changes, particularly affecting the higher GAL1 expression, and may have also exerted indirect regulatory effects on the MAL and IMA gene families. Additionally, the elevated expression of MIG3 plays a critical role in mediating carbon catabolite regulation and the cellular response to ethanol stress (Figure 6). Importantly, although differences in fermentation kinetics could, in principle, influence transcriptomic profiles, this is unlikely to be a major confounding factor in our dataset. Samples were collected at 24 h under comparable biomass conditions (similar OD), corresponding to the exponential phase of growth, and glucose depletion occurred at similar time points across strains. In this context, the observed transcriptional differences are consistent with lineage‐specific strategies of complex sugar utilisation rather than reflecting shifts in growth phase (Cesur et al. 2022; Dong et al. 2017; Villarroel et al. 2021). These regulatory shifts suggest that experimental evolution may partially resolve interspecific cis–trans incompatibilities by reinforcing coherent carbon regulatory hierarchies within the chimeric genome. These adaptations are crucial in high‐gravity fermentations, where sequential sugar consumption (glucose, maltose and then maltotriose) governs overall fermentation efficiency (Alves et al. 2008; Vidgren and Londesborough 2012). Proposed model of the main mechanisms affected in HB44‐II after evolution. In addition, multiple SNPs and SVs targeted genes involved in membrane composition, consistent with their established roles in ethanol tolerance (Albillos‐Arenal et al. 2025; Lairón‐Peris et al. 2021). Interestingly, we observe a large number of SVs across subgenomes. SV events can lead to rapid and significant phenotypic changes via diverse mechanisms, including the modification of open reading frames, changes in gene expression due to copy number variation (CNV), or even the formation of novel genes (O'Donnell et al. 2023). Although the frequency of SNPs is usually higher than that of SVs, the larger size of SVs has been associated with major differences in genome structure, gene expression and protein functionality (Alonge et al. 2020). In this sense, gene expression differences between ancestral and evolved hybrids can also be influenced by differences in SVs spread across the genome, which may partially explain the overlap observed between RNA‐seq and genomic changes in our evolved lineages. This is consistent with previous studies demonstrating that SVs can exert widespread and context‐dependent effects on gene expression and regulatory networks (Alonge et al. 2020; Jeffares et al. 2017). Recent studies in S. cerevisiae have shown that SVs account for a greater proportion of phenotypic variation within populations than SNPs (O'Donnell et al. 2023; Peter et al. 2018). In particular, copy number variations (CNVs) within subtelomeric regions have been linked to enhanced growth under various environmental conditions, including elevated temperature (e.g., PAU gene family), antifungal exposure (ADH4, HXK2) and shifts in carbon sources, such as growth on glucose, fructose, maltose and maltotriose (HXT and MAL gene cluster) (Crandall et al. 2024; Cubillos et al. 2011; Lin and Li 2011; Molinet et al. 2022). Notably, in our study, the S. cerevisiae subgenome exhibited a higher burden of functionally relevant mutations and structural alterations than the S. eubayanus subgenome, suggesting that adaptive evolution preferentially exploited the regulatory and metabolic plasticity of the S. cerevisiae genomic background. Importantly, the evolved isolates analysed here do not represent a uniform genotype, as independent lineages have accumulated distinct combinations of SNPs and SVs. This indicates that these hybrids comprise dynamic and heterogeneous genetic backgrounds rather than fixed clonal entities and thus remain amenable to further rounds of selection and genetic improvement. This heterogeneity may be further accentuated by differences in ploidy across hybrid backgrounds, as the triploid HB44‐II likely buffers the phenotypic effects of deleterious mutations by providing additional gene copies, thereby relaxing purifying selection and enabling the accumulation of a broader spectrum of genetic variation, in contrast to the more constrained diploid HB3‐I background. Taken together, these results suggest that evolvability in synthetic yeast hybrids might be context‐dependent and appears to be linked to the specific genetic background of the parental strains used in this study. The divergent adaptive trajectories observed in HB3‐I and HB44‐II closely mirror the ecological histories of their parental strains. Brewing‐derived hybrids such as HB3‐I exhibited moderate but consistent improvements in maltotriose utilisation, accompanied by molecular changes consistent with fine‐tuning of specific sugar transport systems, reflecting prior optimisation for maltose‐ and maltotriose‐rich brewing environments. In contrast, bioethanol‐derived hybrids such as HB44‐II achieved substantial gains in CO2 production and maltose‐maltotriose metabolism and displayed broader regulatory rewiring, likely reflecting ancestral adaptation to more variable and dynamic carbon sources. Together, these findings suggest that hybrid adaptation does not follow a uniform path but instead exploits lineage‐specific genomic and regulatory potential shaped by parental ecological specialisation. In conclusion, experimental evolution in high‐gravity wort and ethanol stress can significantly enhance the fermentative performance of S. cerevisiae × S. eubayanus hybrids. Adaptive outcomes are influenced by parental background, with beer‐derived hybrids favouring transporter upregulation, while bioethanol‐derived hybrids evolve broader regulatory rewiring. These findings provide a framework for the domestication of hybrid yeasts, enabling tailored improvements aligned with industrial fermentation goals. Author Contributions Tomás A. Peña: investigation, methodology, visualization, software, data curation. Fernanda Jeria: investigation, methodology, formal analysis. Sonia Albillos‐Arenal: writing – original draft, methodology, writing – review and editing, formal analysis, data curation. Pablo Villarreal: funding acquisition, writing – review and editing, investigation, software, formal analysis, data curation, supervision. Darío González‐Venegas: investigation, methodology, visualization, software, formal analysis, data curation. Vasni Zavaleta: investigation, methodology, writing – review and editing, formal analysis. Francisco A. Cubillos: conceptualization, investigation, funding acquisition, writing – original draft, writing – review and editing, formal analysis, project administration, resources, supervision. Funding This work was supported by Agencia Nacional de Investigación y Desarrollo (FONDECYT 1220026, FONDECYT 11240649, ICN17_022, NCN2024_040, FB210008). Universidad de Santiago de Chile, 022443CR_Postdoc. Conflicts of Interest The authors declare no conflicts of interest. Supporting information Figure S1: Experimental evolution scheme for the ancestral hybrid strains, yielding three independent evolutionary lines (I, II, III). Fig S2: Phenotypic profiles of ancestral and evolved hybrids, represented by μmax values normalised to the commercial W34/70 strain and grouped by S. cerevisiae lineage (beer or ethanol). Fig S3: Sequencing depth analysis and ploidy validation of ancestral and evolved hybrid strains. Average read depth per chromosome for lineages (A) HB3 and (B) HB44. The data compares ancestral strains (HB3A/HB44A) with their respective evolved lines (HB3I/HB44I), confirming the maintenance of subgenome ratios across all chromosomes. Fig S4: Number and type of structural variants detected with MUM&Co. Table S1: Parental strains of the hybrids used in this study. Table S2: Hybrids performed by Zavaleta et al. (2024), that were used in this study to evaluate their performance in high‐gravity fermentations. Table S3a: Fermentative capacity of hybrids under high gravity beer wort. R = replicates, SD = standard deviation. Table S3b: Statistics of the fermentative capacity of hybrids under high gravity beer wort compared to the parentals and the commercial strain using Tukey's HSD post hoc comparisons. Table S3c: Pairwise t‐test comparisons of CO2 production between W34/70 and S. cerevisiae parental strains. Table S4: Summary of statistical tests comparing fermentative performance of each hybrid group with that of their corresponding parental strains using an analysis of variance (ANOVA) followed by TukeyHSD. Table S5: Growth rates of the ancestral strains and the evolved lines under different stress conditions. Table S6a: Growth rates measurement (μmax) of ancestral (G0) and evolved lines across fermentation conditions. Table S6b: Post hoc pairwise comparisons of lineages at G250. Table S7a: Total CO2 loss of the evolved lines after 250 generations (G250), and the ancestral hybrids (G0) under high gravity beer wort (20°P, 12°C), and its statistical analysis by using an analysis of variance (ANOVA). Table S7b: Relative CO2 loss of the evolved lines relative to its ancestral strain under high gravity beer wort (20°P, 12°C). Table S7c: Statistics of the groups: One‐way ANOVA and post hoc comparisons of CO2 production across S. cerevisiae and S. eubayanus lineages. Table S8a: Residual maltose, maltotriose and glucose (g/L) during fermentation in malt extract 20°P, 12°C. Table S8b: Final consumption calculations of maltose and maltotriose (g/L) during fermentation in malt extract 20°P, 12°C. Table S8c: Maltose and maltotriose statistics from the residual measurements based on Welch's two‐sample t‐tests comparing ancestral and evolved strains. Table S9: SNPs identified in evolved HB3 and HB44 clones. Alleles across three independently sequenced clones (c1–c3) are shown, with a category indicating whether variants are common (3/3), shared (2/3; clone combination indicated), or unique (1/3; clone indicated). SnpEff gene and impact were extracted from the first ANN annotation. Variants were retained if supported by DP ≥ 10 in ≥ 1 clone carrying the variant, and variants within high‐depth regions (500‐bp windows with depth≥max (40×, 1.5× mean depth), union across clones) or dense variant clusters (≥ 3 variants within a 300‐bp window; all variants in flagged intervals removed) were excluded. Table S10a: BUSCO v5.4.7 stats for hybrid genome assemblies of ancestral and evolved strains using the lineage dataset saccharomycetes_odb10. Table S10b: Structural variants identified in the evolved hybrids when compared with the ancestral hybrids using MUM&Co. Table S11a: Significantlly differentially (p.adj < 0.1) expressed gene (DEG) analysis of HB3, performed in R using the DESeq2 package v4.1.2. All log2 fold changes (log2FC) are presented; values with an absolute log2FC > 1; < −1 are highlighted in bold. Table S11b: GSEA Molecular Function enrichment analysis of HB3‐I versus HB3‐A. Table S12a: Significantlly differentially (p.adj < 0.1) expressed gene (DEG) analysis of HB44, performed in R using the DESeq2 package v4.1.2. All log2 fold changes (log2FC) are presented; values with an absolute log2FC > 1; < −1 are highlighted in bold. Table S12b: GSEA Molecular Function enrichment analysis of HB44‐II versus HB44‐A. Acknowledgements This research was funded by Agencia Nacional de Investigación y Desarrollo (ANID) FONDECYT (1220026), FONDECYT INICIACIÓN (11240649) and ANID ‐ Programa Iniciativa Científica Milenio ICN17_022, and NCN2024_040. FC is funded by FONDECYT 1220026, V.Z. by ANID grant N° 21201566, and P.V. is supported by FONDECYT INICIACIÓN grant N° 11240649. Financial support is also acknowledged to Centro Ciencia & Vida, FB210008, Financiamiento Basal para Centros Científicos y Tecnológicos de Excelencia de ANID. SA acknowledges Universidad de Santiago de Chile POSTDOC_DICYT, Código 022443CR_Postdoc Vicerrectoría de Investigación, Innovación y Creación. Data Availability Statement All raw sequencing data produced in this study, including whole‐genome sequencing (Illumina and Oxford Nanopore) and RNA‐seq reads, have been uploaded to the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) under BioProject ID PRJNA1433379. References Associated Data Supplementary Materials Figure S1: Experimental evolution scheme for the ancestral hybrid strains, yielding three independent evolutionary lines (I, II, III). Fig S2: Phenotypic profiles of ancestral and evolved hybrids, represented by μmax values normalised to the commercial W34/70 strain and grouped by S. cerevisiae lineage (beer or ethanol). Fig S3: Sequencing depth analysis and ploidy validation of ancestral and evolved hybrid strains. Average read depth per chromosome for lineages (A) HB3 and (B) HB44. The data compares ancestral strains (HB3A/HB44A) with their respective evolved lines (HB3I/HB44I), confirming the maintenance of subgenome ratios across all chromosomes. Fig S4: Number and type of structural variants detected with MUM&Co. Table S1: Parental strains of the hybrids used in this study. Table S2: Hybrids performed by Zavaleta et al. (2024), that were used in this study to evaluate their performance in high‐gravity fermentations. Table S3a: Fermentative capacity of hybrids under high gravity beer wort. R = replicates, SD = standard deviation. Table S3b: Statistics of the fermentative capacity of hybrids under high gravity beer wort compared to the parentals and the commercial strain using Tukey's HSD post hoc comparisons. Table S3c: Pairwise t‐test comparisons of CO2 production between W34/70 and S. cerevisiae parental strains. Table S4: Summary of statistical tests comparing fermentative performance of each hybrid group with that of their corresponding parental strains using an analysis of variance (ANOVA) followed by TukeyHSD. Table S5: Growth rates of the ancestral strains and the evolved lines under different stress conditions. Table S6a: Growth rates measurement (μmax) of ancestral (G0) and evolved lines across fermentation conditions. Table S6b: Post hoc pairwise comparisons of lineages at G250. Table S7a: Total CO2 loss of the evolved lines after 250 generations (G250), and the ancestral hybrids (G0) under high gravity beer wort (20°P, 12°C), and its statistical analysis by using an analysis of variance (ANOVA). Table S7b: Relative CO2 loss of the evolved lines relative to its ancestral strain under high gravity beer wort (20°P, 12°C). Table S7c: Statistics of the groups: One‐way ANOVA and post hoc comparisons of CO2 production across S. cerevisiae and S. eubayanus lineages. Table S8a: Residual maltose, maltotriose and glucose (g/L) during fermentation in malt extract 20°P, 12°C. Table S8b: Final consumption calculations of maltose and maltotriose (g/L) during fermentation in malt extract 20°P, 12°C. Table S8c: Maltose and maltotriose statistics from the residual measurements based on Welch's two‐sample t‐tests comparing ancestral and evolved strains. Table S9: SNPs identified in evolved HB3 and HB44 clones. Alleles across three independently sequenced clones (c1–c3) are shown, with a category indicating whether variants are common (3/3), shared (2/3; clone combination indicated), or unique (1/3; clone indicated). SnpEff gene and impact were extracted from the first ANN annotation. Variants were retained if supported by DP ≥ 10 in ≥ 1 clone carrying the variant, and variants within high‐depth regions (500‐bp windows with depth≥max (40×, 1.5× mean depth), union across clones) or dense variant clusters (≥ 3 variants within a 300‐bp window; all variants in flagged intervals removed) were excluded. Table S10a: BUSCO v5.4.7 stats for hybrid genome assemblies of ancestral and evolved strains using the lineage dataset saccharomycetes_odb10. Table S10b: Structural variants identified in the evolved hybrids when compared with the ancestral hybrids using MUM&Co. Table S11a: Significantlly differentially (p.adj < 0.1) expressed gene (DEG) analysis of HB3, performed in R using the DESeq2 package v4.1.2. All log2 fold changes (log2FC) are presented; values with an absolute log2FC > 1; < −1 are highlighted in bold. Table S11b: GSEA Molecular Function enrichment analysis of HB3‐I versus HB3‐A. Table S12a: Significantlly differentially (p.adj < 0.1) expressed gene (DEG) analysis of HB44, performed in R using the DESeq2 package v4.1.2. All log2 fold changes (log2FC) are presented; values with an absolute log2FC > 1; < −1 are highlighted in bold. Table S12b: GSEA Molecular Function enrichment analysis of HB44‐II versus HB44‐A. Data Availability Statement All raw sequencing data produced in this study, including whole‐genome sequencing (Illumina and Oxford Nanopore) and RNA‐seq reads, have been uploaded to the National Center for Biotechnology Information (NCBI) Sequence Read Archive (SRA) under BioProject ID PRJNA1433379.