Deciphering the genetic basis of chlorophyll fluorescence parameters and vegetation indices under heat stress in blueberry

Abstract

Background

Blueberry (Vaccinium section Cyanococcus) is a commercially important fruit crop. Its cultivation is challenging due to specific climate and soil requirements, with even minor temperature fluctuations potentially affecting growth and fruit yield. Heat stress impairs growth and photosynthesis by causing damage to the photosystem II (PSII) complex. Hence, identifying genomic regions influencing the maximum quantum efficiency of PSII and the overall photosynthetic capacity of plants under heat stress is crucial.

Results

In this study, we exposed 266 interspecific cross derivatives of blueberry plants to heat stress in controlled conditions for two consecutive years and evaluated their responses through phenotypic, chlorophyll fluorescence, and vegetation indices measurements. We observed continuous and significant variation for all the measured traits, indicating their quantitative nature. Genome-wide association studies (GWAS) using 126,816 single-nucleotide polymorphisms revealed several candidate genes encoding molecular chaperones, serine-threonine kinases, and DEAD-box proteins, which play critical roles in heat stress responses in plants. Additionally, we identified 10,393 differentially expressed genes (DEGs) from the transcriptomic analysis of V. corymbosum and V. darrowii plants subjected to 6 and 9 hours of heat stress. RNA-Seq data were further confirmed by quantitative real-time PCR analysis of randomly selected genes. Functional characterization through Gene Ontology and pathway analysis of 325 genes commonly expressed in all four comparison groups indicated that DEGs were mainly enriched in protein processing in the endoplasmic reticulum. Through GWAS and transcriptomic analysis, we uncovered 23 genes on chromosomes 1, 2, 3, 4, 5, 7, 8, 9, 10, and 12 associated with chlorophyll fluorescence parameters and vegetation indices under heat stress.

Conclusions

This study provides a deeper understanding of the genetic basis of chlorophyll fluorescence under heat stress in blueberries, offering valuable information for breeding strategies to develop blueberry cultivars with enhanced photosynthetic efficiency.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12870-025-07550-2.

Background

Highbush blueberry (Vaccinium corymbosum section Cyanococcus) is a perennial shrub native to North America and grown for its edible fruits. Its production worldwide, primarily in South America, New Zealand, Australia, and Europe, is increasing. Blueberry fruits are rich in phytochemicals such as anthocyanins, phenolic acids, flavonoids, stilbenes, and tannins. They also possess many nutritive compounds, including sugars, vitamins, carotenoids, and minerals [1, 2] that may have various health benefits [3]. The high content of anthocyanins is linked to improved night vision, brain function, anti-cancer activity, and reduced risk of cardiovascular diseases [4–7]. Hence, blueberries are considered one of the “superfoods.”

Recent changes in climate, mainly driven by global warming, pose significant threats to plant growth and food security [8]. Blueberries are traditionally cultivated in the temperate regions of Canada and the northern United States, but they have expanded to a diverse array of geographical locations across the globe. While these regions offer the ideal climates for blueberry cultivation, maintaining optimal temperatures is crucial. Blueberry plants are particularly sensitive to temperature changes, with an optimal range of 20 °C (night-time) to 25 °C (day-time), with an upper day-time limit of 30 °C [9]. Exceeding this range, especially during flowering, can reduce plant survival and lower fruit quality [10, 11]. Studies have demonstrated that high temperatures during the flowering phase can adversely impact pollination and fertilization in rabbiteye (V. ashei) blueberry cultivar ‘Brightwell’ and some highbush blueberry cultivars [12, 13]. This heightened sensitivity to heat stress has become increasingly concerning as global warming intensifies.

Understanding the ability of plants to tolerate higher temperatures is essential for developing new heat-tolerant blueberry cultivars. Plants respond to heat stress by undergoing several morpho-anatomical transformations influencing their growth and development [14]. Phenotypes such as thinner leaves, spongy layers, and reduced size of mesophyll cells help plants to alleviate the effects of heat stress [11]. The adaptability of blueberry plants to heat stress conditions varies among genotypes. V. darrowii, a native blueberry species in the hot and dry areas of Florida, grows better in warmer climates [15, 16]. V. darrowii plants possess small, parallel-oriented, waxy leaves that help regulate transpiration [17]. Florida 4B (V. darrowii; PI 554904), a wild blueberry with exceptional quality, has been widely introgressed with northern highbush blueberry (NHB) cultivars to develop southern highbush blueberry (SHB) cultivars. Hence, SHBs are considered to be better adapted to heat stress than the NHBs [18].

In addition to the morpho-anatomical adaptations, physiological changes (photosynthesis and respiration) are highly associated with heat stress because temperature profoundly affects net photosynthetic rates. Growth temperatures above optimal temperature during developmental stages can reduce photosynthesis by disrupting chloroplast structure and damaging photosystem Ⅱ (PSII) function [19]. These findings suggest that stomatal characteristics are crucial in regulating transpiration and alleviating the harmful effects of heat stress on chloroplast and membrane stability. Heat-tolerant cultivars exhibited intact chloroplasts and showed increases in electrolyte leakage as well as chlorophyll fluorescence parameters [11]. The maximum efficiency of PSII (Φ_PSII) was found to decrease with increases in heat stress in plants such as cotton (Gossypium hirsutum L.), [20]. Φ_PSII is highly sensitive to temperature increases and is routinely used to phenotype plant cultivars for heat tolerance [21, 22]. Compared to heat-sensitive cultivars, heat-tolerant blueberry cultivars improved their heat dispersing capacity by regulating stomatal traits, thereby protecting chloroplasts from the harmful effects of heat stress [11]. These findings suggest that chlorophyll fluorescence measurements can be used to understand the photosynthetic responses of blueberry plants to heat stress.

In the present study, we evaluated 266 interspecific cross derivatives for heat stress tolerance by measuring the chlorophyll fluorescence parameters, and conducted a genome-wide association study (GWAS) to identify the single-nucleotide polymorphisms (SNPs) and candidate genes regulating chlorophyll fluorescence and vegetation indices under heat stress in blueberry. We further analyzed the previously sequenced transcriptome data of two blueberry species, V. darrowii (a heat-tolerant diploid accession from Florida) and V. corymbosum (a heat-sensitive diploid accession from New Jersey), subjected to heat stress, and uncovered candidate genes by integrating GWAS with differentially expressed genes (DEGs).

Materials and methods

Plant materials

This study used interspecific cross derivatives developed by Dr. Nicholi Vorsa at the Philip E. Marucci Center for Blueberry and Cranberry Research and Extension, Chatsworth, NJ, USA. To develop these cross derivatives, V. darrowii (NJ88-14-3, NJ88-12–41) and V. corymbosum var. caesariense (NJOPB-8, NJOPB-15) were used as grandparents (Fig. S1). NJOPB-15 and NJOPB-8 were collected from an Indigenous population in Burlington County, NJ (39.71° N, 74.51° W), whereas NJ88-12–41 and NJ88-14-03 were collected in 1988 from indigenously grown populations in Liberty County, Florida (30.24° N, 85.01° W) and along Route 98S in the Saint Joseph Bay area of Florida (29.78° N, 85.28° W), respectively [23]. Since both of them are wild collections, they are probably heterozygous in nature. The NJOPB-8 (♀) was crossed with NJ88-12–41 (♂) to produce the hybrid BNJ05-237-8, and NJ88-14-3 (♀) was crossed with NJOPB-15 (♂) to produce the hybrid BNJ05-218-9. Thus, BNJ05-237-8 was in V. corymbosum cytoplasm, and BNJ05-218-9 was in V. darrowii cytoplasm. Finally, these two hybrids were crossed to produce about 1,025 cross-derivatives. From these, we selected a subset of the 260 plants and assessed them for heat stress tolerance, along with six parental lines.

Heat stress evaluation

The 266 cross-derivatives were subjected to heat stress for four days in the growth chamber, which was set at the temperature 40/34 °C (day/night) [19], the photoperiod of 16 h/d (light/dark), and the relative humidity of 50%. The light intensity was approximately 500 µmol m−2 s−1. The phenotype measurements were performed after 4 days of heat stress. Leaf scorch score (LSS) for each plant was visually scored based on the percentage of leaves showing chlorosis and necrosis using a 0–4 scale, where 0 = no chlorosis; 1 = slight chlorosis (1 to 25%); 2 = moderate (26 to 50%) 3 = highly moderate chlorosis (51 to 75%); and 4 = severe chlorosis or dead (76 to 100%).

Chlorophyll fluorescence parameters, quantum yield (Φ_PSII), and rapid fluorescence transient assay [OJIP]) were recorded using FluorPen FP 110 (Qubit Systems, Inc., Canada). The OJIP phase takes less than one second to capture the rapid fluorescence transient after illumination of a dark-adapted leaf sample by high-intensity continuous light [24]. Several studies have concluded that the photosynthetic physiological state of the leaf sample regulates the shape of the OJIP transient momentarily [24, 25]. For the measurement of Φ_PSII and OJIP, three leaves, randomly chosen from each plant, were prepared for dark adaptation using leaf clips. Dark adaptation allows PSII electron acceptors to fully re-oxidize and release any previous quenching, making the measurement possible [26]. It establishes a baseline of minimum fluorescence (Fo) using a low intensity modulated light, and then a saturating light pulse measures maximum fluorescence to calculate the maximum quantum yield of PSII, a key indicator of plant health and stress. Thus, the leaves were dark-adapted for at least 20 minutes for each plant, following the standard protocol outlined in the manual for FluorPen FP 110 (Qubit Systems, Inc., Canada). Measurements were made facing the adaxial surface of the fully expanded leaves, but still attached to the plant. Phenotype data for seven chlorophyll fluorescence parameters were used for GWAS. These traits were, Φ_PSII or termed as quantum yield, actual yield of PSII; Fv/Fo, the efficiency of the oxygen-evolving complex; Fv/Fm, the maximum photochemical efficiency of PSII; ΦPo, the maximum quantum yield of primary PSII photochemistry; PI_ABS, performance index on absorption basis; ABS/RC, light energy absorbed per reaction center; and, E_To/RC, electron transport flux further than plastoquinone A.

Vegetation indices, namely normalized difference vegetation index (NDVI), simple ratio (SR), and greenness (G) indices were recorded using Polypen RP 400 (Qubit Systems, Inc., Canada). The average of three replications is used for the analysis of each of the parameters.

Statistical analysis

The heat stress experiments were conducted in 2019 and 2020. All the data parameters are represented as means ± standard error of three replicates for each plant sample. The phenotype data from both years were combined, and the mean was calculated for each plant parameter and used as the phenotype data in GWAS. Statistical analyses for estimating broad-sense heritability value (h²) and the analysis of variance (ANOVA) were conducted using IBM SPSS version 29.0 (SPSS Inc., Chicago, IL, United States). Duncan’s multiple range test at a 0.05 probability level was used as a posthoc test to assess the significant (p < 0.05) differences among the years.

DNA isolation

For DNA isolation, young leaves of blueberry plants were sampled in 2019 in dry ice and stored at −80 °C. Approximately 100 mg frozen leaf tissue of each sample was homogenized in a 2 ml polypropylene Eppendorf safe-lock microcentrifuge tubes (US Scientific, USA) using TissueLyser II (Qiagen, USA) at a frequency of 30 cycles/s. DNA extraction was carried out using a DNeasy mini plant kit (Qiagen, USA) according to the manufacturer’s instructions [15, 27]. The DNA concentration and quality were examined using a NanoDrop^TM 2000 spectrophotometer (Thermo Scientific, Wilmington, United States) and agarose gel (1.0 %) electrophoresis [28]. The integrity and quality of DNA were further assessed using a Qubit 3.0 Fluorometer (Thermo Fisher Scientific, MA, USA) to select good-quality DNA samples for genotyping by sequencing (GBS) library preparation and sequencing.

Genotyping-by-sequencing analysis and sequence alignment

For GBS, high-quality DNA (∼100 ng/μl) was digested using a type II restriction endonuclease (ApeK1), and the digested DNAs were subsequently ligated to the barcoded adapters as described in Kulkarni et al. [15] to construct the adapter-ligated library. Each sample was pooled and amplified in multiplex using Illumina sequencing primers. The quantity and quality of the library were assessed using Qubit 3.0 Fluorometer and Bioanalyzer 2100 (Thermo Fisher Scientific, MA, USA). Next, the library was sequenced using paired-end (PE) reads of 150 bp on the Illumina NextSeq 500 system at the Department of Biology, West Virginia State University, Institute, WV. The Illumina bcl2fastq2 conversion software v2.20 (Illumina, San Diego, CA, USA) was used to demultiplex the sequencing reads and convert the resulting image files in base call (BCL) formats into FASTQ format. The V. corymbosum cv. ‘Draper’ v1.0 genome was used as a reference sequence (http://gigadb.org/dataset/100537) for read mapping. The filtered sequencing reads were mapped to the 12 largest scaffolds of the reference genome. These scaffolds are numbered 1, 2, 4, 6, 7, 11, 12, 13, 17, 20, 21, and 22, and represent 1–12 V. corymbosum chromosomes, respectively [29, 30]. Variant calling was performed using GB-eaSy (https://github.com/dpwickland/GB-easy), which significantly improves the variant calling accuracy of paired-end reads from GBS data [31]. The resulting variant call file (VCF) was used for further filtering and analysis. The SNPs with minor allele frequency (MAF) ≥0.05% and call rate of at least 90% were retained for the analysis.

Haplotype block analysis

We estimated the linkage disequilibrium (LD) and analyzed the average size range of haplotype blocks for all the chromosomes using SVS v8.8.5. (Golden Helix, USA) as given in Kulkarni et al. [15]. Adjacent and pairwise LD measures were estimated individually for all the selected SNPs.

GWAS analysis and identification of candidate genes

Utilizing high-quality SNPs, we conducted GWAS by adopting MLMM developed by the Efficient Mixed-Model Association eXpedited (EMMAX) method, utilized within the software platform SVS v8.1.5. The SNP p-values obtained by GWAS underwent sequential false discovery rate (FDR) determination and Bonferroni correction. Manhattan plots for associated SNPs were visualized in GenomeBrowse v1.0 (Golden Helix, Inc). The percentage of phenotypic variation exhibited by significant SNPs was estimated by the coefficient of determination (R²). The SNPs were annotated using the V. corymbosum cv. ‘Draper’ v1.0 genome sequence [29], and gene predictions were made. Annotations using the Blast2GO tool with BLASTp search against the NCBI non-redundant protein database were performed to predict the gene models and candidate genes flanking the SNPs.

Transcriptome analysis

In this study, we mapped previously sequenced RNA-seq reads [17] to the 12 longest scaffolds of the V. corymbosum cv. ‘Draper’ v1.0 genome sequence [29] using the STAR universal RNA-seq alignment tool with default parameters [32] to generate a BAM alignment. The read count tables for the genes were created using BAM alignment and the general feature format of genome annotation with the HTSeq R package [33]. The counts were normalized by using reads per kilobase of transcripts per million (RPKM). The gene expression based on the read counts was studied using fragments per kilobase of transcripts per million (FPKM). The FPKM values for the identified gene were estimated on the basis of the read count table, the total number of reads per sample, and gene length in kb. The differentially expressed gene counts were obtained using DESeq [34]. DEGs with a cut-off value of log2fold-change (FC) >1 and FDR < 0.05 were selected.

Functional annotation, GO, and pathway analysis

Gene ontology (GO) enrichment analysis was performed with BLAST2GO to determine the biological significance of the DEGs (https://www.blast2go.com/). Significantly enriched GO terms (FDR< 0.05) were displayed using SRPlot. For pathway analysis, identified DEGs were matched to annotation files, and functional descriptions were assigned using the Kyoto Encyclopedia of Genes and Genome (KEGG, http://www.genome.jp/kegg/) database [35]. DEGs aligned to the blueberry cultivar ‘Draper’ genome were matched to the gene IDs given in the NCBI Entrez database.

Quantitative real-time PCR (qRT-PCR)

Total RNA for each sample was extracted using the Spectrum™ Plant Total RNA Kit (Sigma-Aldrich, USA). Three biological replicates for treatment and control samples of each plant species were used. The RNA quality and concentration were checked using a NanoDropTM 2000 spectrophotometer (Thermo Scientific, United States) and agarose gel (1.0 %) electrophoresis using 1X MOPS buffer. Primer 3 software was used to design primers for all the genes. The amplicon size of the primers was set between 100 and 250 bp. The primers were synthesized from Integrated DNA Technologies . (Coralville, IA, USA).

Quantitative gene expression of selected DEGs was analyzed with the Applied Biosystems 7500 Real-Time PCR System (Thermo Fisher Scientific, USA). The selected candidate gene expression was normalized with Actin as an internal reference gene. Three technical replicates were used for each plant sample. The RT-qPCR for the selected genes was performed using Power SYBR™ Green RNA-to-CT™ 1-Step Kit (Thermo Fisher Scientific, USA). The 15 μL reaction mixture contained 7.5 μL 2X Power SYBR™ Green RT‑PCR Mix, 1.5 μL each forward and reverse primers (5 μM), and 100 ng RNA (2 μL). The standard qRT-PCR was performed following the thermal cycling conditions given in the product manual. Relative gene expression and fold change were determined by the 2^−ΔΔCt method by comparing the expression in heat stress samples (6 h and 9 h) with the expression of control samples (0 h) in both species [36, 37]. The primer sequences are given in Table S1.

Results

Phenotypic performance of interspecific cross-derivatives under heat stress conditions

We measured LSS, chlorophyll fluorescence parameters and vegetation indices across 266 interspecific cross derivatives of blueberry to investigate their responses to heat stress. Four days of heat stress significantly affected the plants, as sensitive plants showed leaf scorching symptoms (Fig. S2). Continuous and significant variation was observed for all the measured traits in the interspecific cross-derivatives (Fig. 1, Table S2). Most traits showed transgressive segregation, suggesting that several genes control the traits and are amenable to GWAS.

Fig. 1.

Phenotypic variation observed in the interspecific cross derivatives. A Frequency distribution of the measured chlorophyll fluorescence parameters and vegetation indices in interspecific cross derivatives under heat stress, showing continuous trait expression. B Box plots showing the year-wise range and distribution of the measured phenotypes

The Pearson pairwise correlation analysis revealed strong relationships among the measured chlorophyll fluorescence parameters and associated traits, as well as with LSS (Table 1). LSS negatively correlated with five chlorophyll fluorescence parameters and two vegetation indices, indicating that these parameters are related to heat stress tolerance. The correlation coefficient (r) values for many relationships were high. Φ_PSII positively correlated with Fv/Fo (r = 0.36), Φ_Po (r = 0.92), and E_To/RC (r = 0.55) but negatively correlated with ABS/RC (r = −0.61). ABS/RC also negatively correlated with Fv/Fo (r = −0.61), Φ_Po (r = −0.74), and PI_ABS (r = −0.36). We observed strong positive correlations between Fv/Fo and Φ_Po (r = 0.58), Fv/Fo and PI_ABS (r = 0.90), Φ_Po and E_To/RC (r = 0.51), and NDVI and SR (r = 0.95). The broad-sense heritability estimates (h²) for Φ_PSII, E_To/RC, ΦPo, ABS/RC, NDVI, SR, and G ranged from 0.33 to 0.65 (Table S3).

Table 1.

Pearson correlation analysis of the measured parameters in the blueberry cross derivatives subjected to heat stress

Parameter	LSS	ΦPSII	Fv/Fo	Fv/Fm	ΦPo	PIABS	ABS/RC	ETo/RC	NDVI	SR
ΦPSII	−0.23*
Fv/Fo	−0.26*	0.68**
Fv/Fm	−0.27*	0.93**	0.77**
ΦPo	−0.27*	0.93**	0.77**	1.00**
PIABS	−0.34**	0.72**	0.70**	0.69**	0.69**
ABS/RC	0.19	−0.76**	−0.56**	−0.66**	−0.66**	−0.56**
ETo/RC	−0.18	0.51**	0.36**	0.46**	0.46**	0.52**	−0.16*
NDVI	−0.21*	0.1	0.07	0.18**	0.18**	0.19**	−0.14*	0.09
SR	−0.22*	0.12*	0.11	0.19**	0.19**	0.21**	−0.15*	0.1	0.95**
G	−0.04	−0.1	0	0.13*	0.13*	−0.14*	0	−0.19**	0.26**	0.13*

^*Correlation is significant at the 0.05 level (2-tailed)

^**Correlation is significant at the 0.01 level (2-tailed)

Genotyping-by-sequencing summary

GBS of 266 plants generated >976 million reads of 75 bp length (Table S4). An average of 3.67 million reads with tags per sample was obtained. The barcoded tags with at least three read counts were used to detect SNPs. The reads were aligned to the chromosome-scale V. corymbosum cv. ‘Draper’ v1.0 genome sequence [29]. An average of 3.08 million reads with a tag per sample were effectively aligned to the reference genome. In total, ~821 million reads (overall mapping rate 84.5%) were mapped to the V. corymbosum reference sequence. The SNPs with minor allele frequency (MAF) < 0.05 and call rate < 0.9 were removed from the analysis. We selected 126,816 SNPs distributed across the 12 chromosomes for further study, averaging 4 SNPs per 1-kb genome length. The number of SNPs mapped to the 12 chromosomes ranged from 8,597 for chromosome 7 (VACCDSCAFF12) to 12,896 for chromosome 1 (VACCDSCAFF1) (Table 2, Fig. 2).

Table 2.

Chromosome-wise summary of the single-nucleotide polymorphism (SNP) statistics from the genotyping-by-sequencing analysis of the interspecific cross derivatives

Chr	Scaffold	Chr Length	Raw SNPs	Filtered SNPs	Average filtered SNPs per kb of genome length
1	VACCDSCAFF1	46,295,995	115,017	12,896	4
2	VACCDSCAFF2	44,818,276	112,676	11,847	4
3	VACCDSCAFF4	42,981,373	98,542	10,374	4
4	VACCDSCAFF6	42,795,824	97,289	9,665	4
5	VACCDSCAFF7	41,705,179	108,082	10,878	4
6	VACCDSCAFF11	40,122,599	98,808	11,009	4
7	VACCDSCAFF12	39,741,682	85,235	8,597	5
8	VACCDSCAFF13	39,652,356	100,130	10,742	4
9	VACCDSCAFF17	38,874,919	92,770	10,437	4
10	VACCDSCAFF20	37,996,905	89,420	9,867	4
11	VACCDSCAFF21	37,975,728	91,709	10,305	4
12	VACCDSCAFF22	37,315,645	88,374	10,199	4
Total			1,178,052	126,816

Fig. 2.

Distribution of single nucleotide polymorphisms (SNPs) on 12 blueberry chromosomes (The horizontal axis shows the chromosome length)

GWAS for chlorophyll fluorescence and associated traits

GWAS involving a multi-locus mixed linear model (MMLM) revealed 965 SNPs significantly associated with Φ_PSII, Fv/Fm, PI_ABS, Φ_Po, Fv/Fo, E_To/RC, and ABS/RC. For vegetation indices NDVI, SR, and G index, 358 SNPs were identified. In total, 1,323 SNPs for chlorophyll fluorescence and associated traits measured in the interspecific cross-derivatives were identified. The maximum proportion of phenotypic variation (R²) explained by the associated SNPs ranged from 5.86% for Fv/Fo to 14.29% for ABS/RC. For vegetation indices, R² values ranged from 4.06% to 7.32%. (Figures 3 and 4, Table S5).

Fig. 3.

A Manhattan plot of genome-wide association analyses for Φ_PSII using a multi-locus mixed linear model (MMLM). B Box plots of the allele effects of 3 SNPs associated with Φ_PSII in interspecific cross derivatives

Fig. 4.

Manhattan plots of genome-wide association analyses for five chlorophyll fluorescence traits using a multi-locus mixed linear model (MMLM)

Φ_PSII is the quantum yield of PSII photochemistry and is used to estimate the overall in vivo photosynthetic capacity [38]. We identified 196 significantly associated SNPs for Φ_PSII, of which 104 clustered on chromosome 3 (Table S5). These SNPs might be in high LD and inherited together as haplotypes. Predominantly, 3 SNPs, VACCDSCAFF4:35919458-SNV (on chromosome 3), VACCDSCAFF12:34659265-SNV (on chromosome 7), and VACCDSCAFF22:1148614-SNV (on chromosome 12) were significantly associated with Φ_PSII (Figure 3). These SNPs encoded HVA22-like protein, DEAD-box ATP-dependent RNA helicase, and putative Hemopexin-like domain-containing protein, respectively, which are reported to be induced by environmental stresses, such as dehydration, salinity, and extreme temperatures [39–42].

The maximum efficiency of the oxygen-evolving complex, represented as Fv/Fo, is considered the most sensitive parameter of plant photosynthetic capacity. We found 32 SNPs explaining 2.4% to 5.9% of phenotypic variation significantly associated with Fv/Fo (Table S5). Genes harboring the SNPs encoded proteins such as pentatricopeptide repeat-containing protein, DnaJ subfamily C member 14-like, and WRKY transcription factor 43.

We identified 281 SNPs significantly associated with multiple chlorophyll fluorescence parameters and vegetation indices (Fig. S3). Genomic regions carrying these SNPs were annotated to identify the candidate genes and their functions. The 6.8 Mb chromosomal region on chromosome 3 carrying 22 SNPs for Φ_PSII contained genes encoding proteins such as phosphate transporter PHO1 homolog 3-like, zinc finger protein-like, cell division cycle-associated 7-like protein, serine/threonine-protein kinase, O-fucosyltransferase 34-like, and putative pentatricopeptide repeat-containing protein Figure S3, Table S5. Another crucial region on chr 12 spanning 3.66 Mb contained 15 SNPs, which encoded proteins such as pentatricopeptide repeat-containing protein, chaperone protein dnaJ 49-like, serine-threonine kinase receptor-associated protein-like isoform X1, and protein MAINTENANCE OF PSII UNDER HIGH LIGHT 1 (Figure S3, Table S5). Besides these two genomic regions, there were other SNPs encoding heat shock proteins (HSP, VACCDSCAFF7:19264268-SNV, VACCDSCAFF22:3008931-SNV) and WRKY transcription factor (VACCDSCAFF22:1422577-SNV). The SNP VACCDSCAFF12:33880097-SNV on chromosome 7 was associated with six chlorophyll fluorescence parameters, namely Φ_PSII, ABS/RC, E_To/RC, Φ_Po, Fv/Fm, and PI_ABS, and explained phenotype variation of 5.08%, 2.95%, 5.33%, 5.08%, 5.08%, and 6.13%, respectively (Table 3, Fig. S4). The candidate gene harboring this SNP encoded Tudor domain-containing protein 3 (Arabidopsis homolog AT5G19950.3). Another SNP VACCDSCAFF20:12840637-SNV on chromosome 10 was also associated with these six parameters, namely, Φ_PSII, ABS/RC, E_To/RC, Φ_Po, Fv/Fm, and PI_ABS, and explained phenotype variation of 4.27%, 5.79%, 4.73%, 5.09%, 5.09%, and 4.63%, respectively Table 3.

Table 3.

Candidate genes and SNP loci significantly (p < 0.001) associated with multiple chlorophyll fluorescence parameters in the interspecific cross derivatives under heat stress

Gene name	Associated SNPs	Scaffold	Chr	Position	R2 range (%)	Chlorophyll fluorescence parameters	Arabidopsis homolog
Hypothetical protein RclHR1_00930026	VACCDSCAFF1:18336214-SNV	1	1	18336214		Fv/Fo, Fv/Fm, ΦPo, PIABS	-
N-glycosylase/DNA lyase OGG1	VACCDSCAFF2:19896806-SNV	2	2	19896806	3.75–4.5.75.5	ΦPSII, ABS/RC, PIABS	AT1G21710.1
	VACCDSCAFF2:19896835-SNV	2	2	19896835	3.84–4.43.84.43	ΦPSII, ABS/RC, PIABS	AT1G21710.1
	VACCDSCAFF2:19896836-SNV	2	2	19896836	3.74–4.45.74.45	ΦPSII, ABS/RC, PIABS	AT1G21710.1
	VACCDSCAFF2:19896903-SNV	2	2	19896903	3.43–5.52.43.52	ΦPSII, ABS/RC, PIABS	AT1G21710.1
HVA22-like protein a	VACCDSCAFF4:35919458-SNV	4	3	35919458	4.09–6.25.09.25	ΦPSII, ABS/RC, PIABS	AT1G74520.1
DnaJ subfamily C member 14-like	VACCDSCAFF7:19264080-SNV	7	5	19264080	4.25–5.27.25.27	ΦPSII, Fv/Fo, ΦPo, PIABS	AT1G79030.1
Tudor domain-containing protein 3	VACCDSCAFF12:33880097-SNV	12	7	33880097	2.95–6.13.95.13	ΦPSII, ABS/RC, ETo/RC, ΦPo, PIABS	AT5G19950.3
Pentatricopeptide repeat-containing protein	VACCDSCAFF12:33949584-SNV	12	7	33949584	3.61–5.04.61.04	ΦPSII, ABS/RC, Fv/Fm, ΦPo	AT5G55740.1
Serine-threonine kinase receptor-associated protein-like isoform X1	VACCDSCAFF12:34074639-SNV	12	7	34074639	4.44–6.32.44.32	ΦPSII, ABS/RC, Fv/Fm, ΦPo	AT1G52730.2
DEAD-box ATP-dependent RNA helicase	VACCDSCAFF12:34659265-SNV	12	7	34659265	4.27–6.32.27.32	ΦPSII, ABS/RC, Fv/Fm, ΦPo	AT1G52730.2
Chaperone protein DNAJ 49-like	VACCDSCAFF12:34718350-SNV	12	7	34718350	3.56–5.57.56.57	ΦPSII, ABS/RC, Fv/Fm, ΦPo	AT3G57340.2
	VACCDSCAFF12:34718648-SNV	12	7	34718648	4.5–7.19.5.19	ΦPSII, ABS/RC, Fv/Fm, ΦPo	AT3G57340.2
	VACCDSCAFF12:34718659-SNV	12	7	34718659	4.14–6.73.14.73	ΦPSII, ABS/RC, Fv/Fm, ΦPo	AT3G57340.2
Otogelin like	VACCDSCAFF20:12840637-SNV	20	10	12840637	4.27–5.79.27.79	ΦPSII, ABS/RC, ETo/RC, ΦPo, PIABS	AT1G13380.1
	VACCDSCAFF22:1148606-SNV	22	12	1148606	3.95–5.07.95.07	ΦPSII, ABS/RC, Fv/Fo, PIABS	AT3G57340.2
Putative Hemopexin-like domain-containing protein	VACCDSCAFF22:1148614-SNV	22	12	1148614	3.79–5.28.79.28	ΦPSII, ABS/RC, Fv/Fo	AT5G41620.1
U11/U12 small nuclear ribonucleoprotein (59 kDa)	VACCDSCAFF22:1283341-SNV	22	12	1283341	3.3–4.85.3.85	ΦPSII, ABS/RC, Fv/Fo	AT2G46200.2

GWAS identified many significant SNPs associated with 3 to 5 chlorophyll fluorescence parameters. These SNPs explained phenotypic variation ranging from 3.56 to 7.19% for the respective traits (Table 3). Several other SNPs showed significant variability for the measured parameters and encoded important genes such as DEAD-box ATP-dependent RNA helicase, pentatricopeptide repeat-containing protein, serine-threonine kinase receptor-associated protein-like isoform X1, chaperone protein dnaJ 49-like, N-glycosylase/DNA lyase OGG1, putative Hemopexin-like domain-containing protein, U11/U12 small nuclear ribonucleoprotein 59 kDa protein-like, and HVA22-like protein.

Transcriptome profiling by RNA sequencing

To identify the differentially expressed heat-responsive genes, we mapped previously sequenced RNA-seq reads [17] to the V. corymbosum cv. ‘Draper’ v1.0 reference genome sequence [29]. We found 4,106 and 6,287 DEGs expressed at 6-h and 9-h stress, respectively, in V. corymbosum and V. darrowii plant samples (Fig. 5). Three hundred twenty-five genes showed differential expression in the four comparison groups. Of these, 257 genes were upregulated in all four groups, whereas 59 were downregulated. Of the remaining nine genes, eight were upregulated in V. corymbosum but downregulated in V. darrowii, whereas one gene, maker-VaccDscaff4-snap-gene-147.72 was downregulated in V. corymbosum but upregulated in V. darrowii (Table S6). Genes encoding RPP13-like protein, known for their involvement in heat stress responses in maize [43], and methyltransferase, which plays an important role in plant metabolism and stress adaptation [44], were the prominent genes differentially expressed between the two blueberry species.

Fig. 5.

Differential gene expression observed in response to heat stress in V. corymbosum and V. darrowii. A-C: Venn diagrams showing the common as well as total DEGs expressed at 6 h and 9 h heat stress in both V. corymbosum and V. darrowii species. D The top 10 gene ontology (GO) terms enriched among the DEGs under heat stress in V. corymbosum and V. darrowii species. E Top 20 enriched KEGG pathways differentially expressed in V. darrowii and V. corymbosum plants under heat stress. The size of the dots represents the number of DEGs mapped to the indicated pathways, whereas the p-value range is represented by color, with extreme red showing the highest lop p-values

Twenty-seven of 325 genes encoded HSPs, which are rapidly induced in response to heat stress in grapes (Vitis vinifera L.) and other plants [45]. The top 20 up- and down-regulated genes between both species are given in Table S7. Thus, our differential gene expression analysis suggests that both the blueberry species adopt different mechanisms to alleviate the effects of heat stress.

The RNA-Seq results were further validated using qRT-PCR. We selected six DEGs for validation, of which five were upregulated genes (maker-VaccDscaff22-snap-gene-107.34, maker-VaccDscaff13-snap-gene-322.19, maker-VaccDscaff13-augustus-gene-52.35, maker-VaccDscaff22-augustus-gene-28.24, and augustus_masked-VaccDscaff6-processed-gene-123.1), and one was downregulated gene (augustus_masked-VaccDscaff2-processed-gene-54.2). The expression patterns shown by these six genes by qRT-PCR followed the same patterns seen in RNA-seq results, and all of them except augustus_masked-VaccDscaff6-processed-gene-123.1 had comparable log2 fold change (Fig. S5).

Functional annotation and classification of DEGs

We performed GO enrichment analysis of 325 DEGs using BLAST2GO according to the categories of biological processes, molecular functions, and cellular components. GO enrichment analysis organized 325 DEGs into 31 biological processes, 17 molecular function, and 21 cellular component terms. The Top 10 GO terms from each category are shown in Fig. 5. We also performed a seperate GO analysis of DEGs in V. corymbosum and V. darrowii at 6 h and 9 h heat stress against the control samples. The highly enriched GO terms included “cellular process”, "biosynthetic process," "protein modification," and "stress response" in the biological process component, "membrane" in the cellular component, and "nucleotide binding," "catalytic activity" "transferase activity", and "binding" in the molecular function component (Fig. S6). The Kyoto Encyclopedia of Genes and Genomes (KEGG) analysis of 325 DEGs assigned 92 genes to 63 pathways significantly enriched in the analysis study. The pathway analysis indicates that many of the DEGs are enriched in "vitamin B6 metabolism", "protein processing in endoplasmic reticulum," "limonene and pinene degradation," and "selenocompound metabolism" (Fig. 5, Figs S7, S8, Table S8) pathways. The most enriched GO terms were "protein processing in the endoplasmic reticulum," "plant-pathogen interactions," and "spliceosome."

Next, we predicted the biological functions associated with the 325 commonly expressed genes using the BLASTx algorithm to compare the sequence queries against the non-redundant protein database at NCBI. The analysis revealed the functions of 308 DEGs, whereas annotation information for 17 DEGs was unavailable. Twenty-four genes encoded heat shock proteins/factors, and three genes each encoded small HSPs, NAC domain-containing proteins, and chaperone proteins (Fig. S9). In addition, several transcription factors like dehydration-responsive element-binding protein 2C-like, auxin response factor 19, ethylene-responsive transcription factor ERF071, MADS-box transcription factor 23 isoform X2, NAC domain-containing protein 72-like, GTE2, and MYB-like were differentially expressed. The findings highlight the distinct transcriptomic responses of these species to heat stress, potentially aiding heat tolerance research.

Combining GWAS with RNA-seq results for candidate gene identification

To pinpoint candidate genes controlling chlorophyll fluorescence parameters and vegetation indices under heat stress, we combined GWAS and RNA-seq results. We selected the top 10 SNPs (based on high phenotypic variation) for each trait (100 SNPs for 10 traits), and only chose DEGs mapped within the distance of the largest LD block size for the respective chromosome (Table 4). For instance, to identify gene models for SNPs on chromosome 12, DEGs mapped within 194 kbp on either side of the SNP were selected. This way, we identified 11 DEGs mapped within the distance of the largest LD blocks (Fig. 6) for each chromosome. We predicted biological functions for these 11 candidates by performing a BLASTP search against the ‘V. corymbosum cv. ‘Draper’ v1.0 genome’ proteins [29]. These genes, located on chromosomes 2, 3, 4, 8, 9, 10, and 12, were predicted to have significant biological functions, such as a receptor-like protein and a transcription termination factor involved in stress responses and plastid development (Table S9). Notably, two NAC transcription factors and genes coding for small HSPs, ethylene-responsive transcription factors, and zinc finger proteins were identified, which are known to play roles in heat stress tolerance. These findings suggest their crucial role in conferring heat tolerance in V. darrowii.

Table 4.

Details of the genes identified through combined genome-wide association and transcriptome analysis

Gene	Chr	Start position	End position	Trait	Distance from the SNP (kb)	Largest LD Block size (kb)	Functional annotation
VaccDscaff2-snap-gene-267.44	2	26,724,956	26,726,221	NDVI	-210.8	425	Transcription termination factor MTERF6, chloroplastic/mitochondrial-like
VaccDscaff4-augustus-gene-93.28	3	9,296,804	9,300,766	Fv/Fm	-21.7	425	Rhomboid-like protein 14, mitochondrial
VaccDscaff4-processed-gene-93.2	3	9,322,227	9,325,671	Fv/Fm	-	425	Receptor-like protein 12
VaccDscaff6-processed-gene-123.1	4	12,328,383	12,329,624	SR	-92.7	418	Ethylene-responsive transcription factor ERF054-like
VaccDscaff13-augustus-gene-24.20	8	2,465,456	2,468,582	SR	-70.6	261	Protein OPI10 homolog
VaccDscaff13-augustus-gene-52.35	8	5,259,424	5,263,977	ABS/RC	37.5	261	Zinc finger protein 511
VaccDscaff13-snap-gene-322.19	8	32,248,523	32,249,662	SR	-43.7	261	E3 ubiquitin-protein ligase ATL42-like
VaccDscaff17-augustus-gene-331.40	9	33,150,482	33,153,445	ABS/RC	58.6	243	NAC domain-containing protein 72-like
VaccDscaff17-augustus-gene-331.41	9	33,170,705	33,171,852	ABS/RC	78.8	243	NAC domain-containing protein 72-like
VaccDscaff20-processed-gene-127.33	10	12,751,584	12,754,015	NDVI	92.9	255	3-ketoacyl-CoA synthase 6
VaccDscaff22-augustus-gene-28.24	12	2,813,459	2,815,662	SR	10.0	194	Small heat shock protein, chloroplastic-like isoform X2

Fig. 6.

Heatmap based on the Log2FC values of 11 differentially expressed genes commonly identified in the genome wide association study and transcriptome analysis (Red: upregulated, Blue: downregulated)

Discussion

With climate change and rising global temperatures, blueberry cultivation faces significant challenges. Understanding genetic and molecular mechanisms regulating plants' response to abiotic stresses such as heat stress is crucial to enhancing thermotolerance in blueberry cultivars. Heat stress reduces the ability of plants' PSII reaction centers to capture and use light energy [46]. Chlorophyll fluorescence measurements provide a non-destructive means of assessing plant health quickly. Φ_PSII and OJIP are sensitive indicators of the functionality of PSII in plants, and provide insights into the effect of heat stress on the photosynthetic structure and photosystem activity. In this study, we evaluated 266 interspecific cross derivatives for heat stress tolerance and measured chlorophyll fluorescence parameters and vegetation indices to provide insights into the genetic basis of photosynthetic responses to heat stress conditions. The population used here is widely expected to segregate for heat stress tolerance, as it is derived from an interspecific cross between diploid V. corymbosum (from New Jersey) and southern diploid V. darrowii (from Florida) species [23]. We found significant phenotypic variation for chlorophyll fluorescence-related traits, including Fv/Fo and Fv/Fm in blueberry cross derivatives. We observed a significant decrease in parameters such as Fv/Fo and Fv/Fm in some blueberry plants, indicating that heat stress inhibits the light energy conversion efficiency of PSII in blueberry plants. Such plants showed higher leaf scorching symptoms after 4 days of heat stress. The continuous variation and transgressive segregation observed in most traits indicate that this population could be used to investigate the genetic and molecular mechanisms regulating heat-stress tolerance blueberry. To our knowledge, this is the first study investigating the genetic dissection of chlorophyll fluorescence parameters under heat stress in blueberry.

Our GWAS identified significant associations for chlorophyll fluorescence and related traits in blueberries under heat stress, contributing valuable insights into the genetic regulation of this complex trait. Heat tolerance is controlled by multiple quantitative trait loci (QTL), and pinpointing the candidate genes has remained a challenge. Through this study, we identified 1,323 SNPs, serving as QTLs, and explaining 5.86–14.29% of the phenotypic variation for chlorophyll fluorescence and associated traits, a number that compares favorably with similar research in other crops [47–51]. Two key SNPs emerged from the analysis: VACCDSCAFF12:33880097-SNV on chromosome 12, associated with Tudor domain-containing protein 3, and VACCDSCAFF20:12840637-SNV on chromosome 20, linked to a domain of unknown function (DUF1218) (Table 2). Tudor domain proteins are known for their role in chromatin structure, gene regulation, and association with stress adaptation [52–54]. DUF1218, although not fully characterized, has been implicated in cell wall biology and responses to abiotic and biotic stress [55]. Other significant SNPs were associated with genes encoding DnaJ subfamily C member 14-like protein and chaperone protein dnaJ 49-like protein, which are critical in maintaining protein homeostasis under stress conditions [56]. Furthermore, DEAD-box RNA helicases, identified on chromosomes 1, 2, 12, and 22, are key players in unwinding RNA and regulating protein remodeling, essential processes in managing plant stress responses [57]. These findings collectively suggest that the identified SNPs and associated genes are involved in various mechanisms that enable blueberry plants to tolerate heat stress, paving the way for further research into their functional roles in stress adaptation.

Blueberry plants have evolved various mechanisms to cope with heat stress, an essential trait for maintaining yield amidst unpredictable climate conditions [10]. Understanding these adaptive mechanisms is crucial, especially as commercial blueberry varieties exhibit phenotypic variation in heat stress tolerance influenced by their genetic heritage [15]. Comprehensive investigations into the morpho-physiological, molecular, and genetic responses to heat stress are warranted. Transcriptome analysis reveals the genes and pathways regulating heat stress responses [58]. In this regard, we identified several DEGs critical for heat stress adaptation. Heat stress triggers the production of HSPs and chaperones regulated by heat stress transcription factors, stabilizing protein conformation and assisting in proper protein folding [59, 60]. We observed the upregulation of 27 HSPs/molecular chaperones in two blueberry genotypes, including high- and low-molecular-weight HSPs. Additionally, 18 transcription factors (TFs) were upregulated or downregulated in response to heat treatment in V. darrowii and V. corymbosum, highlighting their role in regulating stress-responsive genes [61, 62].

Further GO and pathway analysis revealed that heat stress activates various cellular mechanisms related to membrane integrity and protein processing. Heat stress can damage cellular structures, leading to increased membrane fluidity and protein denaturation [63, 64]. Our GO analysis indicated that many DEGs are linked to membrane-related components, suggesting their role in mitigating stress damage [17]. The endoplasmic reticulum (ER) plays a crucial role in responding to heat stress by processing misfolded proteins. A higher expression of HSPs is seen when plants are exposed to heat stress, which shows that HSPs are an important component of plant adaptation to heat stress [65]. We found that 31 genes, most of which encode heat shock proteins involved in ER pathway were upregulated in both the species (Table S8). These genes encoded heat shock proteins, luminal-binding proteins (BiP), dnaJ protein, endoplasmin homolog (also known as GRP94), calnexin homolog, protein disulfide isomerase-like 2–3, derlin-1 isoform, which are mainly involved in calcium regulation, protein folding and modification, protein translocation, and protein degradation [66]. This study shows that heat treatment activates ER pathway genes, underscoring their importance in plant adaptation [17, 67]. Besides the ER pathway, Vitamin B6 has recently been implicated in plant defense against oxidative stress [68]. In our study, three genes from the Vitamin B6 metabolism pathway encoding pyridoxal 5'-phosphate synthase (PDX1.2) and phosphoserine aminotransferase were upregulated [68, 69].

Although GWAS have helped discover the genetic variants associated with complex traits in blueberry [30, 70–73], the resolution of GWAS is usually insufficient to map the causal genes, especially in plants with long LD decay distances [70, 74, 75]. To refine and pinpoint the trait-associated gene/s, GWAS needs to be combined with linkage mapping or transcriptome analysis [76] as demonstrated in several recent studies identifying potential candidate genes governing abiotic stress tolerance [77–83]. In this study, the combination of GWAS and RNA-seq data greatly helped to narrow down the target regions and pinpointed 11 potential candidate genes regulating heat stress responses in blueberry (Table 3). These genes encode proteins critical in plant stress response and adaptation. For instance, the gene VaccDscaff4-processed-gene-93.2, which encodes Receptor-like protein 12, was downregulated in both blueberry species after heat stress [84, 85]. Another gene, VaccDscaff2-snap-gene-267.44, encodes Transcription Termination Factor MTERF6, which is crucial for chloroplast and mitochondrial functions, highlighting its role in abiotic stress response [86, 87]. Furthermore, NAC domain protein genes VaccDscaff17-augustus-gene-331.40 and VaccDscaff17-augustus-gene-331.41 were upregulated following heat stress, linking them to abiotic stress tolerance [88, 89]. Overall, our findings provide significant insights into the genetic basis of chlorophyll fluorescence under heat stress in blueberries, identifying potential targets for developing cultivars with improved photosynthesis in warmer climates. Different methods, such as transgenics, gene knockout, overexpression, and CRISPR/Cas9-mediated gene silencing, can be explored for functional validation of these genes for their future application in blueberry breeding programs.

Conclusions

This study advances our understanding of the genetic mechanisms underlying plants' response to heat stress. Through GWAS combined with transcriptome analysis, we identified 23 key genes (encoding known heat-stress related proteins such as heat shock protein, DEAD-box helicase, Zinc finger protein, Ethylene-responsive transcription factor, and NAC domain containing protein) and pathways (vitamin B6 metabolism, and protein processing in the endoplasmic reticulum) that play critical roles in the plant's response to heat stress. The upregulation of HSPs and transcription factors, alongside the involvement of membrane-related components and endoplasmic reticulum pathways, highlights the complex network of responses that enable blueberries to withstand elevated temperatures. These findings contribute to the existing knowledge of plant stress physiology and pave the way for future research to enhance the photosynthetic efficiency of blueberry cultivars.

Abbreviations

GWAS

Genome-wide association studies

QTL

Quantitative trait locus

SNP

Single nucleotide polymorphism

HSP

Heat shock protein

ER

Endoplasmic reticulum

DEG

Differentially expressed gene

TF

Transcription factors

PSII

Photosystem II

NDVI

Normalized difference vegetation index

SR

Simple ratio index

G

Greenness index.

MAF

Minor allele frequency

NHB

Northern highbush blueberry

SHB

Southern highbush blueberry

MMLM

Multi-locus mixed linear model

GO

Gene ontology

KEGG

Kyoto Encyclopedia of Genes and Genomes

qRT-PCR

Quantitative real-time polymerase chain reaction

GBS

Genotyping-by-sequencing

LD

Linkage disequilibrium

Authors’ contributions

K.P.K. conceived and designed the experiments and wrote the main manuscript. J.J.C. and N.V. provided plant material and reviewed the manuscript; P.N., K.P.K., and U.K.R. performed GBS, post-sequencing, and GWAS analyses; M.I. provided the blueberry genome sequence and reviewed the manuscript; J.C. and S.E. contributed critical comments to the draft. K.M. and U.K.R. conceived the project and reviewed the manuscript. All authors reviewed and approved the final version of the manuscript.

Funding

We thank the United States Department of Agriculture (USDA-NIFA) for funding (nos. 2018-38821-27744, 2018-67014-27622, and 2023-38821-39921).

Data availability

The raw paired-end Illumina sequencing reads generated in the current study are available in the Sequence Read Archive (SRA) at NCBI under the Bio project accession number PRJNA687760.

Declarations

Ethics approval and consent to participate

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Consent for publication

All authors consent for publication.

Competing interests

The authors declare no competing interests.