Mature fig trees (Ficus carica L., cv. Brown Turkey) grown in a commercial orchard near Beer Tuvia, Israel (31°44′13.66″N, 34°43′32.05″E) were used for two field experiments conducted in May 2012 and 2013. The Brown Turkey cultivar belongs to the ‘common fig’ type that produces parthenocarpic fruit. All trees were maintained according to commercial-production practices with short pruning trimming up to biennial branches such that the braking buds are 2 years old.

Fruit corresponding to growth phase I (15 mm in diameter, n = 2,000), positioned sixth or seventh on the shoot, were tagged on the trees; 1,000 fruit, covered with 100-mesh bags to prevent natural pollination by wasps, served as the non-pollinated control population, while the other 1,000 were hand-pollinated as follows: several ‘caprifig’-type fig fruits of selected varieties were harvested at the ripening phase. Fruits were cross-sectioned to reveal the stamens. The back of every half was gently tapped to shake the pollen into a small vial which was immediately stored at -20°C. Pollen (0.5 g) was dissolved in 100 mL 2% sucrose solution (2 g sucrose in 100 mL ddH₂O) and injected into the fruit through the ostiole with a plastic syringe. To ensure that the entire inflorescence comes into contact with the pollination solution, fruit were filled until a drop of the pollination solution came out through the ostiole. The pollination solution volume injected into each individual fruit thus varied with the size of the inner fruit cavity. Following hand-pollination, all fruits were covered with 100-mesh bags to prevent natural pollination. Fruits from each treatment were collected every 7 days for 90 days post-pollination to produce a developmental profile. At 90 days after pollination, all ripened fruit were harvested and classified according to their ripening stage, which was determined by the percentage of the red color coverage of the outer fruit skin (Figure 1).

For the high-throughput sequencing, three biological replicates were collected from pools of four fruit at each developmental stage that were dissected into pulp and inflorescence and immediately frozen in liquid N₂, pulverized and stored at -80°C. For the hormone analysis, 15 whole fruits from each developmental stage were pooled and immediately frozen in liquid N₂, pulverized, lyophilized and stored at -20°C.

About 50 mg of two lyophilized biological replicates from each sample were prepared for hormone analysis. Quantification of ABA, ABA metabolites, indoleacetic acid (IAA), IAA metabolites, gibberellins, gibberellin metabolites and cytokinins was conducted at the National Research Council of Canada (Saskatoon, SK, Canada) according to their published protocols¹.

Total RNA was isolated from pulp (peel and receptacle) and inflorescence according to Jaakola et al. (2001). RNA concentration was determined in a NanoDrop ND-1000 spectrophotometer and its integrity was checked by running 1 μL in a 1% (w/v) agarose gel stained with Bromophenol blue.

Total RNA was extracted from 40 tissue samples. RNA fragmentation, double-stranded cDNA synthesis and adaptor ligation were performed using the Truseq^TM RNA Sample Prep Kit-v2 (Illumina, San Diego, CA, USA) according to the manufacturer’s instructions. Library concentration and size were assayed using the Qubit^® dsDNA HS Assay Kit in a Qubit^® 2.0 Fluorometer (Invitrogen, Lidingo, Sweden) and the Agilent DNA1000 Kit on a 2100 tape station (Agilent Technologies, Nærum, Denmark). Each library was normalized to a final concentration of 100 nM. All 22 samples were grouped into pools and diluted to a final concentration of 10 nM. Libraries were sequenced on a genome analyzer equipped with a paired-end module (Illumina) at the Technion – Israel Institute of Technology (Haifa, Israel) to generate 100-bp paired-end reads.

Raw reads were subjected to a filtering and cleaning procedure using the FASTX Toolkit ² (version 0.0.13.2) as follows: (i) read-end nucleotides with quality scores <30 were trimmed using the fastq_quality_trimmer; (ii) reads with less than 70% base pairs with quality score ≤30 were removed using the fastq_quality_filter. A total of ∼1.9 billion cleaned reads, obtained after processing and cleaning, were assembled de novo using Trinity software [version trinityrnaseq_ r20131110 (Grabherr et al., 2011); 25 mer k-mer size]. Filtering of the likely contig artifacts was carried out as follows: (i) abundance estimates were calculated for each contig using the RSEM software; (ii) only contigs representing more than 1% of the per-component (IsoPct) expression level were retained. The resulting de-novo assembly-generated transcriptome catalog is shown in Table 1.

The transcriptome datasets are available in the NCBI Sequence Read Archive (SRA) under BioProject accession PRJNA322124.

The resulting contigs were annotated using the Basic Local Alignment Search Tool (BLASTX) (Altschul et al., 1990) against the NCBI non-redundant (Nr) and Malus × domestica protein (from the genome database for Rosaceae) that chosen for being a fleshy climacteric fruit like the Ficus carica and for being well annotated in the NCBI, databases with an E-value cut-off of 10^-5 and TAIR. Blast2GO software (Conesa et al., 2005) was used for gene ontology (GO) classification³.

Transcript quantification (number of reads per gene) from the RNA-Seq data was performed using the bowtie aligner (Langmead et al., 2009) and the expectation-maximization method (RSEM), which handles read-mapping uncertainty with a statistical model by estimating maximum-likelihood expression levels (Li and Dewey, 2011). Differential expression analysis was performed with the edgeR package (Robinson et al., 2010). Transcripts with more than twofold differential expression levels with false discovery-corrected statistical significance of at most 0.001 were considered differentially expressed (Benjamini and Hochberg, 1995). The expression patterns of the transcripts in the different samples were studied using cluster analysis of the differentially expressed transcripts in at least one pairwise sample comparison. Then, the Trinity protocol (Haas et al., 2013) expression normalization was designed using trimmed mean of M-values (TMM), following fragments per feature kilobase per million reads mapped (FPKM) calculations. Hierarchical clustering of gene expression and visualization of heat maps were performed using R Bioconductor (Gentleman et al., 2004). We used the tool at http://bioinformatics.psb.ugent.be/webtools/Venn/ for Venn diagram construction.

Gene ontology-enrichment analysis was carried out using the Blast2GO (Conesa et al., 2005) program based on Fisher’s Exact Test (Upton, 1992) with multiple testing correction of false discovery rate (FDR) (Benjamini and Hochberg, 1995). The threshold was set as a FDR with corrected P-value of less than 0.05. GO analysis was performed by comparing the GO terms in the test sample to those in a background reference. The REVIGO web server (Supek et al., 2011) was used for reduction and visualization of the GO terms. A hypergeometric test using a MATLAB script was performed for the detection of significantly enriched pathways in a gene list with a cut-off P-value < 0.05.

To establish the transcriptome of ripening parthenocarpic and pollinated fig fruit, 40 cDNA libraries from fig fruit pulp and inflorescence at five developmental stages: (i) 8 weeks after pollination (8WAP), (ii) 9WAP, both from fruit developmental phase II, (iii) 10% fruit ripening, where 10% of the fruit surface turned to red (iv) 60% fruit ripening, where 60% of the fruit surface turned to red and (v) 100% fruit ripening, from fruit developmental phase III, were created (Figure 1). Illumina HiSeq 2000 was used for sequencing, yielding 803.4 million 100-bp paired-end reads with an average of 20 million reads per sample. Properties of the de novo assembly after quality trimming and filtration are shown in Table 1. For gene annotation, the assembled transcript datasets were compared with the NCBI and Malus × domestica protein databases using the BLASTX program (Altschul et al., 1990), resulting in 53,938 (36.7%) and 49,128 (33.4%) annotated genes for parthenocarpic and pollinated fig fruit, respectively. GO classification via BLAST2GO revealed high representation of cellular metabolic processes, primary metabolic processes, organic substance metabolic processes, biosynthetic processes and macromolecular metabolic processes in the parental category. In addition a high representation of binding (including ion, protein, heterocyclic and organic cyclic compound binding), hydrolase, transferase and oxidoreductase activities in the molecular function parental category, and high representation of cell, intracellular, cytoplasm, organelle and membrane-bound organelle in the cellular component category, were confirmed (Supplementary Figure S1).

To evaluate gene expression during fruit development for each of the four different treated tissues: (i) parthenocarpic inflorescence, (ii) pollinated inflorescence, (iii) parthenocarpic pulp and (iv) pollinated pulp, differentially expressed transcripts were evaluated using 10 pair wise comparisons (all vs. all, edgeR, FDR < 0.001, >2 fold, Figure 2A). Hierarchical clustering analysis of gene representations in parthenocarpic and pollinated inflorescence and pulp revealed four distinct cluster patterns for each of the four different tissues (Supplementary Figure S2). The gene-representation expression patterns in the different hierarchical clusters clearly distinguished between developmental phases II and III and could be classified into two major groups: genes whose expression declines when entering ripening phase III and genes that have increased levels at this developmental stage (Supplementary Figure S2). The distinct gene-expression behavior between phases II and III allowed us to look further at the genes that are activated in each phase (phase II – 8WAP and 9WAP, and phase III – 10, 60, and 100% ripening) as one group for each phase.

For each hierarchical cluster, GO-enrichment analysis was performed to shed light on the most prominent molecular functions and biological processes during ripening of the specific parthenocarpic or pollinated tissue. While the number of downregulated and upregulated enriched molecular functions was similar in the pollinated inflorescence, the parthenocarpic inflorescence showed a similar number of downregulated genes but three times higher enrichment in upregulated genes. The major upregulated enriched genes in the parthenocarpic inflorescence were involved in the cellular response to internal and external signals and protein metabolism (Supplementary Table S1). While the number of upregulated genes in the parthenocarpic pulp was similar to that in the parthenocarpic inflorescence, the pollinated pulp showed almost three times higher enrichment compare to the pollinated inflorescence (Supplementary Table S1). In addition to the cellular responses and protein metabolism, the pollinated pulp showed upregulation and enrichment in hormone-related gene expression (Supplementary Table S1).

To evaluate the effect of pollination on fig fruit development and ripening, a set of differentially expressed genes in either pulp or inflorescence at each developmental stage was mined. All of these genes showed a minimum twofold higher FPKM with FDR threshold of 0.05 (edgeR) (Figure 2).

In general, the number of differentially expressed genes in pollinated vs. parthenocarpic inflorescences was higher than the number in pollinated vs. parthenocarpic pulp (Figure 2). Except for the 10% ripened pulp sample, the number of upregulated genes in all pollinated samples was higher than in their parthenocarpic counterparts (Figure 2). Hierarchical cluster analysis of gene-expression patterns revealed four significantly different groups of up-represented genes during phase II (parthenocarpic inflorescence and pulp, pollinated inflorescence and pulp) and two groups in phase III: a major group of up-represented genes in the pollinated inflorescences and a small group in the pollinated pulp (Figure 2B). To calculate the intersections of molecular function lists that share higher representation in all parthenocarpic or pollinated phase II or III stages, we used the public web tool http://bioinformatics.psb.ugent.be/webtools/Venn/ to create a Venn diagram and list for each group (Figure 2C, Table 2). In general, the number of up-represented genes throughout phases II and III in the inflorescence was higher than in the pulp, while a larger number of up-represented genes was found in phase II compared to phase III (Table 2). The number of up-represented genes in the pollinated phase II inflorescence was ca. double that in the pollinated phase III inflorescence and three times that in the parthenocarpic phase II inflorescence (Table 2). The number of up-represented genes in pollinated phase II pulp was four times lower than that in the pollinated phase II inflorescence but ca. three times higher than that in parthenocarpic phase II pulp (Table 2). The up-represented genes in the pollinated phase III inflorescence were the only substantial up-representation in any of the phase III samples (Table 2). The molecular functions color, taste and aroma, storage, seeds and embryo, chlorophyll, and hormone-related function genes had a relatively high number of up-represented genes in the pollinated inflorescence over the parthenocarpic inflorescence during developmental phases II and III (Table 2). No other significantly up-represented genes in one tissue or treatment over the other could be seen in phase III. In phase II, the parthenocarpic inflorescence showed up-representation of protein degradation and cell response genes relative to the pollinated inflorescence. In the phase II pulp tissue, the pollinated fruits showed up-representation of storage, protein degradation and cell response-related genes relative to the parthenocarpic pulp, while the parthenocarpic pulp showed up-representation of cell response-related genes relative to the pollinated pulp (Table 2).

In this study, for the first time, a ripening-related hormotome revealed the levels of ABA, IAA, GA, cytokinins and their derivatives during the ripening of parthenocarpic and pollinated fig fruit (Figure 3, Supplementary Figure S3). Whole-fruit ABA content increased during fruit ripening with a maximum of 4,170 ng/g DW in pollinated 30% ripened fruit compared to a maximum of 2,016 ng/g DW in parthenocarpic 10% ripened fruit, and declined toward the 100% ripe stage with ABA contents of 1,743 and 1,577 ng/g DW in pollinated and parthenocarpic fruit, respectively (Figure 3A). The prominent ABA-degradation pathway seemed to be via the ABA–GE conjugate as its concentration was three to four times higher in both parthenocarpic and pollinated fruit than the concentration of ABA, showing a maximum of 6,182 ng/g DW in the 30% ripened parthenocarpic fruit and 19,507 ng/g DW in the pollinated 60% ripened fruit, while no decline in ABA–GE was detected in either case (Figure 3A). The second ABA-degradation pathway that goes through PA toward DPA was one order of magnitude lower than the ABA content. While the PA content showed no differences between parthenocarpic and pollinated fruit, with a maximum 50 and 125 ng/g DW in the 30% ripened fruit, respectively, the DPA content in the pollinated fruit reached a maximum 221 ng/g DW at the 30% ripening stage, two times higher than the maximum content of DPA, 137 ng/g DW, in parthenocarpic fruit at the 60% ripening point (Supplementary Figure S3A). Endogenous auxin, which was represented by the biologically active IAA and its non-active conjugate with aspartic acid IAA–Asp was detected in all examined samples (Figures 3B,C). While IAA content increased throughout the entire ripening period, showing no significant differences between parthenocarpic and pollinated fruit, the IAA–Asp conjugate maintained constant levels in all 10–100% ripened parthenocarpic fruit with peaks at the 30 and 60% ripening stages of 168 ng/g DW (Figure 3C).

The only active GA during the ripening process was GA8, which is the GA1 inactivator. GA8 content showed no differences between parthenocarpic and pollinated fruit throughout the entire ripening process, displaying some decline from 53 ng/g DW and 71 ng/g DW 8WAP to 23 ng/g DW and 23 ng/g DW at the 100% ripening stage, respectively (Figure 3D). The non-active GA19, GA20, and GA21 showed low but steady content throughout the ripening process, from the 10 to 100% ripening stage (Supplementary Figure S3B).

No active cytokinin was detected in either parthenocarpic or pollinated fruit during the ripening process. However, both cis and trans Z catabolites, c-ZOG and t-ZOG, were detected during fruit ripening (Supplementary Figure S3C) as parthenocarpic fruit showed a maximum concentration of 98 ng/g DW t-ZOG at 8WAP, followed by low levels during later ripening stages. Pollinated fruit showed significantly higher t-ZOG content throughout the entire ripening process, from 10 to 100% ripening, with a maximum of 177 ng/g DW at the 10% ripening stage followed by a constant decline (Supplementary Figure S3C). c-ZOG content in both parthenocarpic and pollinated fruit declined from 25 ng/g DW and 18 ng/g DW at the 10% ripening stage to 10 ng/g DW and 7 ng/g DW at the 100% ripening stage, respectively (Supplementary Figure S3C).

To better understand the significant ripening-related concentrations of ABA (Seo and Koshiba, 2011), auxin (Mano and Nemoto, 2012) and ethylene (Owino et al., 2006; Freiman et al., 2014), we analyzed the transcript expression of each hormone-biosynthesis and catabolism gene in our fig transcriptome. The genes transcripts were tblastn vs., our database and cross validate with NCBI and Malus. Thirteen ABA-related transcripts were identified (Supplementary Figure S4). The ABA-synthesis NCED, ABA2 and AAO transcripts showed the highest expression levels at the 10% ripening and 8WAP stages, (Figures 4A–H). The ABA-degradation ABA-8′-h gene showed later expression peaks in both pulp and inflorescence tissues at 60 and 100% ripening stages, respectively, while the alternative degradation UGT gene, transcript displayed a maximum expression level at 8WAP (Figures 4I–L). In general, differentially expressed transcripts showed higher expression levels in pollinated fruit than in parthenocarpic ones throughout the ripening process, particularly in the inflorescence (Figure 4). Both NCED1 and 2 showed higher expression levels in parthenocarpic fruit inflorescence and pulp at 9WAP, before ripening onset, and higher expression levels in the pollinated fruit inflorescence at the 10% ripening stage (Figures 4A–D). At this stage, the NCED1 and 2 genes were higher in the pollinated inflorescence, while NCED1 remained higher throughout ripening (Figures 4A–D). NCED2 was higher in the pollinated pulp stages at 10–100% ripening, while NCED1 was higher in the pollinated pulp from the 60% ripening stage onward (Figures 4B,D). ABA2 showed an early ripening peak at 8–9WAP with higher expression levels in pollinated vs. parthenocarpic inflorescences, followed by lower levels during the ripening stages, as shown for UGT and AAO, which were upregulated in the pollinated pulp compared to the parthenocarpic one Figures 4E–H,K,L).

A total of 19 auxin-biosynthesis- and catabolism-related gene transcripts were identified (Figure 5; Supplementary Figures S5). TDC1, YUCCAL, CYP71A13, TAAL and GH3.1 transcripts (Figure 5) and IAA content showed increased expression and higher concentrations simultaneously, throughout fruit ripening (Figure 3). The first gene involved in IAA metabolism, TDC1, showed higher expression in the pollinated fruit inflorescence over the parthenocarpic ones with maximum representation 9WAP followed by a gradual decreasing throughout ripening. In the pulp TDC1 showed higher representation at the 9WAP point in the parthenocarpic fruit over the pollinated ones (Figures 5A,B). YUCCAL expression was 10-fold higher in the inflorescence than in the pulp, with a constant increase throughout the ripening process and no significant differences between pollinated and parthenocarpic fruit. This gene, however, was upregulated in parthenocarpic pulp at both 60 and 100% ripening stages (Figures 5C,D). CYP71A13L had low expression levels in the green 8–9WAP fruit, and was generally higher in parthenocarpic than in pollinated ones. Expression levels of this gene were 10-fold higher during ripening as parthenocarpic fruit maintained their relatively higher transcript levels compared to pollinated fruit (Figures 5E,F). Low expression of TAAL, with relatively higher representation in the parthenocarpic fruit (Figures 5G,H), was also found when the auxin conjugation gene GH3.1 showed a peak at the 10% ripening stage, with higher representation in the pollinated inflorescence. GH3.1 expression was higher in pollinated pulp compared to parthenocarpic pulp during the earlier stages, while showing a significant increase at the later ripening stages in the latter (Figures 5I,J).

Despite a previous study by Rosianski et al. (2016) showing no difference in ethylene production between parthenocarpic and pollinated fig fruit, two transcripts among the 14 ethylene-biosynthesis-related genes, SAM synthase and SAM2 transcripts (Figures 6A,B, Supplementary Figure S6) showed higher expression levels in pollinated vs. parthenocarpic inflorescences during the ripening process, while SAM3 had higher levels in parthenocarpic and pollinated fruit at 8–9WAP and 10–100% ripening stages, respectively (Figures 6C,D). ACS2 and 4 showed extremely low expression levels in both pulp and inflorescence tissues at 8–9WAP and high levels at the 10–100% ripening stages. ACS2 expression was generally higher in the pollinated inflorescence, in contrast to ACS4 which was highly expressed in the parthenocarpic inflorescence (Figures 6E–H). ACOL3 showed similar expression levels in pollinated and parthenocarpic fruit throughout fruit development with a significant expression peak at 100% ripening in parthenocarpic fruit (Figures 6I,J).

The well-studied strong connection between seed and hormone metabolism led us to produce a comprehensive hormotome of the five common plant hormones: ethylene, ABA, auxin, GA, cytokinin and their derivatives from the onset of, and throughout ripening of the ‘common’ fig fruit cv. Brown Turkey. A schematic model of hormone production during the ripening of parthenocarpic and pollinated fruit is provided in Figure 7.

Ethylene production was not affected by the pollination event, as both pollinated and parthenocarpic fruit showed a typical ethylene burst at the onset of fruit ripening with no significant difference between them (Rosianski et al., 2016) (Figure 7). The peak of ABA production at 30% ripening (Figures 3A and 7) was in agreement with its production in tomato fruit (Srivastava and Handa, 2005), where the importance of the reproductive tissue in hormone production was seen in parthenocarpic tomatoes, which had a lower proportion of ABA than their seeded counterparts (Sjut and Bangerth, 1982). Similarly, the maximum ABA concentration in the pollinated fig fruit was doubles that in parthenocarpic ones (Figure 3A). In addition, ABA production was higher in the pollinated fruit, expressed by higher production of its PA, DPA and ABA–GE derivatives, than in the parthenocarpic fruit (Figures 3A and 7, Supplementary Figure S3).

IAA production in both pollinated and parthenocarpic ‘Brown Turkey’ initiated in phase II and remained high during the entire ripening phase III (Figures 3B and 7). In the climacteric apple, IAA concentration showed a three- to fourfold increase prior to the rapid rise in ethylene concentration but fell to its original level as this rise occurred (Mousdale and Knee, 1981). The same pattern of early-growth-stage IAA production before the onset of chlorophyll degradation was reported in the non-climacteric strawberry fruit (Symons et al., 2012). In tomato, the rise and fall of IAA production also occurs prior to the ethylene peak but appears much earlier, before the onset of chlorophyll degradation at the ‘breaker’ stage when cell expansion and fruit ripening take place (Srivastava and Handa, 2005). Comparison of pollinated vs. parthenocarpic tomato fruit revealed higher production of IAA in the seeded pollinated fruit than in the parthenocarpic ones (Sjut and Bangerth, 1982). IAA production that precedes the ethylene peak at ripening was also reported by Lodhi et al. (1969) during first parthenocarpic and second pollinated crop’s phase II of ‘San-Pedro’ fig fruit, in addition to a later peak in IAA production during growth and ripening phase III. These two peaks of IAA production differ from our current results (Figures 3B and 7), possibly due to the different fig cultivars or the difference in fig types—‘San-Pedro’ vs. ‘common’ type fig.

No active cytokinins or gibberellins were found in either parthenocarpic or pollinated fig fruit hormotomes. This finding is in agreement with other climacteric fresh fruit such as tomato, where GA and cytokinin contents decrease before ripening onset (Srivastava and Handa, 2005). An exception to this pattern was GA8, which was strongly induced during phase III, presumably to inhibit GA1 synthesis (Figure 3D).

The higher representation of transcripts and their differential expression in the inflorescence as compared to the pulp may suggest a major role for the reproductive inflorescence tissue in controlling fruit development and ripening. In addition, the seeds in the “true” fruit contributed to the ripening, while gene expression in the vegetative pulp tissue was generally lower. This general conclusion is in agreement with previous work on ethylene control of fig fruit ripening where ethylene production was mainly controlled in the fruit inflorescence (Freiman et al., 2015). Here, two ethylene-metabolism genes were differentially represented in the fig pulp: ACS4 and ACOL3. A higher representation of yet another ethylene-metabolic gene, ACS2, was observed in the pollinated inflorescence, in contrast to ACS4 which was expression was higher in the parthenocarpic inflorescence, suggesting an effect of seed production on ethylene metabolism.

The ABA-production pattern during most climacteric fruits’ ripening usually precedes or parallels ethylene production (Crane and Blondeau, 1949; Lara and Vendrell, 2000; Zhang et al., 2009; Symons et al., 2012; Leng et al., 2014). For example, in the climacteric tomato fruit, ABA content increases at the cell expansion and fruit ripening and decreases before the ripening phase, which includes loss of chlorophyll color and red lycopene accumulation with no change in fruit size (Srivastava and Handa, 2005). It was shown that this peak in ABA content precedes ethylene production in tomato seeds and flesh by ∼4 days (Zhang et al., 2009; Leng et al., 2014). Similarly, the ABA peak was shown to precede ethylene production during the climacteric persimmon fruit’s ripening by 30 days, and by 10–20 days in the climacteric apple fruit (Lara and Vendrell, 2000; Leng et al., 2014). The unique fig fruit exhibited similar characteristics, with elevated ABA production at the beginning of the ripening period in parthenocarpic and pollinated fruit (Figure 3A). ABA production paralleled the onset of ethylene production which peaked 1 day later, at 30% ripening. In contrast, climacteric fig fruit also possess characteristics that can be related to non-climacteric fruit (Sozzi et al., 2005; Owino et al., 2006; Freiman et al., 2014). In our study, the increase in ABA content is not only seems to activate the ripening process as in climacteric fruit, but also coincided with increased fruit size as in non-climacteric strawberry (Symons et al., 2012). The increase in ABA content started at the end of the quiescence phase II throughout maturation phase III (fruit size 40–55 mm). The advanced ABA production in the pollinated vs. parthenocarpic fruit was reflected in the representations of ABA-metabolism genes, including the key gene in ABA production, NCED (Figures 3A and 4). The importance of the fig inflorescence NCED1 and 2 transcripts, which showed up-representation before ACS2 and 4, is in agreement with tomato fruit, where ABA induces ethylene biosynthesis via the regulation of ACS and ACO expression (Zhang et al., 2009).

The pattern of TDC1, CYP73A13 and TAAL auxin-metabolism gene transcripts in the fig transcriptome was in agreement with the profile of auxin content in the fig hormotome, which increased during fruit ripening with no general difference for pollinated vs. parthenocarpic fruit (Figures 3, 5 and 7). In contrast, higher representation in the pollinated fruit inflorescence compared to its parthenocarpic counterpart was seen for GH3.1, the gene responsible for IAA–Asp conjugation (Figures 5I,J). The non-active IAA–Asp conjugate, suggested to directly affect responses to abiotic stress, was found at higher levels in the pollinated vs. parthenocarpic fruit.