Habanero pepper seedlings (C. chinense Jacq. ‘Jaguar’) with ≈45 d of germination were used. They were transplanted into polyethylene bags of 12 kg of capacity containing BS or RS. There were 15 seedlings for each type of soil. The seedlings were cultivated in greenhouse conditions for 150 days post-transplant (DPT). Information on the average conditions of temperature, relative humidity, and intense light inside the greenhouse during the 5 months of the experiment is in Table S1. The values are averages of the house means of one of the measured parameters. Plants were distributed in blocks of three plants by type of soil, alternating blocks of soil type. The characteristics of the soils used for the experiments are described extensively in Oney-Montalvo et al. [28].

Plants were irrigated with 0.5–1.5 L of water. They were fertilized using Triple-18 fertilizer (0.7713 g/plant) two days per week. Also, micronutrients were applied to leaves, one day per week, using Bayfolan Forte. In the floral period, a growth regulator containing phytohormones (Biozyme) was applied every 15 d.

Flowers were labeled when 50% of the plants had started flowering period. Flowers were labeled on anthesis day during 60 d and the flowers and fruit numbers were daily registered. The total fruit number per plant and the total number of fruits of 45 days post-anthesis (DAA) per plant were determined. Also, the fruits of 45 DAA were divided into four categories according to their length (1–1.99 cm, 2–2.99 cm, 3–3.99 cm, and ≥4 cm), to analyze in detail the effect of the soil type in the growth of the fruits, and the total number of fruits per plant, total fresh weight of fruits per plant and average weigh of fruit per plant were determined in each category. ANOVA was carried out using Tukey analysis (p ≤ 0.05) with n = 8 plants for the analysis of number of flowers and total fruits, n = 13 plants for the flower abortion, and n = 7 plants for the 45 DAA fruit analysis.

Fruits of 25, 45, and 60 DAA were frozen in liquid nitrogen and stored at −80°C. These times were selected due to what has been observed that corresponds to different development stages of the fruits and have different behavior in the changes of the capsaicinoid content in habanero pepper [9]. We wanted to compare the content of capsaicinoids between soils and to analyze if there are significant differences that could be explored in the transcriptomic analysis. Two pools of five fruits randomly selected for each maturation stage and soil type were ground with a blender in liquid nitrogen, and a lyophilized sample (0.15 g) was used to extract the capsaicinoids in 20 mL anhydride ethanol according to Córdova-Alvarado et al. [29], resuspending the extracts in ethanol grade HPLC. The capsaicinoids were determined by thin-layer chromatography (TLC) according to Córdova-Alvarado et al. [29] by loading 5 μL of resuspended extract as a band of 5.0 mm length with a Linomat 5 equip (CAMAG) from a distance of 10 mm above the bottom edge of the plates, and developing the chromatography without solvent saturation until reaching 7.0 cm from the bottom of the plates. Then the plates were set in a CAMAG TLC Scanner 4 equipment and the capsaicinoids were quantified by densitometry at 282 nm. For the extracts of fruits of 25 and 45 DAA, the determination was made in ten samples (n = 10), and for the extracts of fruits of 60 DAA, the determination was made in twelve samples (n = 12). In parallel, standards of capsaicin were used in a range of 0.25 to 3.0 µg to construct a calibration curve. For the statistical analysis, two-way ANOVA was carried out using Tukey analysis (p ≤ 0.05) with the factors: soil type (BS and RS) and development stage of the fruit (25, 45, and 60 DAA) with n = 10 samples in fruits of 25 and 45 DAA, and n = 12 samples in fruits of 60 DAA.

We wanted to evaluate the effect of the soil type in the expression of genes of the capsaicinoid synthesis pathway, we selected the fruits of 25 DAA (intermediate development stage) because it has been reported that the fruits in intermediate development stages are very active in the capsaicinoid synthesis, and the genes of the corresponded pathway have high levels of expression [15,22]. Two pools of five fruits of 25 DAA of each type of soil were used to extract the RNA to obtain two biological replicates per soil type (BS1 and BS2 for black soil and RS1 and RS2 for red soil. The fruits, previously grounded, were prepared to RNA-Seq using the Illumina sequencing platform with the services of Macrogen (South Korea).

Total RNA was extracted (0.1 g) using the kit RNeasy Plant Minikit (Qiagen, USA) following the manufacturer’s instructions. The RNA extracted was treated with the kit Turbo DNase (Invitrogen, USA) following the manufacturer’s instructions. RNA integrity was verified by electrophoresis in agarose gel (1.5%) in TAE 1X. RNA was quantified in a NanoDrop 2000 spectrophotometer (ThermoScientific, USA). The RNA was dried with the kit RNAstable/RNAstable LD (Biomatrica, USA) following the manufacturer protocol stored in bags with silica gel (Biomatrica, USA), and then sent to Macrogen for sequencing. The RNA integrity number (RIN) was acceptable for all the samples (8.7 for BS1 and BS2 and 8.2 for RS1 and RS2).

The four RNA samples were prepared for RNA-Seq using the Illumina TruSeq Stranded mRNA LT Sample Prep Kit for library construction of each biological replicate (BS1, BS2, RS1, and RS2), obtaining a total of four libraries. Briefly, the mRNA was purified from the total RNA using poly-T oligo-attached magnetic beads from each biological replicate. The mRNA was randomly fragmented, and then the cleaved RNA fragments were reverse transcribed to obtain the first cDNA strands using random hexamer primers. Then, the second cDNA strands were synthesized using RNase H and DNA Polymerase I. Adapters were ligated onto both ends of the cDNA fragments preparing them for hybridization onto the flow cell. Later, the fragments were amplified by PCR, and those with insert sizes between 200–400 bp were sequenced with the Illumina platform from both ends to obtain 101 nt paired ends reads. The raw reads were deposited in the NCBI database (BioProject under the accession number PRJNA983842).

The raw reads were filtered using the software Trimmomatic v.0.38 [30] to eliminate adapter sequences and bases with a quality score of Q < 3 from the ends of the fragments. Also, we used a sliding window of four bases and a mean quality of Q15 to trim bases of low quality from the reads. Afterward, reads with lengths shorter than 36 bp were dropped. With these parameters, trimmed reads of high quality were obtained in the four libraries (Table 1). Statistics of the quality of the raw and trimmed data were obtained with the software FastQC v.0.11.7 [31]. Assembly of the trimmed data was performed using the software Trinity (version trinityrnaseq_r20140717) with default parameters [32] obtaining an acceptable assembled transcriptome statistic (Table 2A). Remapping of the trimmed reads to the assembled contigs was performed using the program Bowtie v.1.1.2 [33]. Finally, the contigs were filtered and clustered into non-redundant representative transcripts (unigenes) using the software CD-HIT v.4.6 [34].

The prediction of the open reading frames (ORFs) of at least 100 aa in length was made using the program TransDecoder v.3.0.1 [35] to identify candidate coding regions in the unigenes library.

To estimate the abundance of the unigenes in each sample (BS1, BS2, RS1, and RS2), trimmed reads of each sample were aligned to the assembled unigenes library with Bowtie v.1.1.2, and then, the abundance of the unigenes was calculated in each sample using the software RSEM v.1.3.1 [36], expressing the result in read counts, fragments per kilobase of transcript per million mapped reads (FPKM) and transcripts per million (TPM).

The functional annotation of the unigenes library was performed against the public databases of Gene Ontology (GO, version v20180319), Kyoto Encyclopedia of Genes and Genomes (KEGG KO_EUK, version 20220103), and against the nucleotide (NT, version 20180116) and non-redundant protein (NR, version 20180503) databases of the NCBI. For functional annotation of unigenes on the NT database of the NCBI, we carried out BLASTN (BLAST v.2.9.0) with an e-value cut-off of 1.0e-5. For functional annotation on the NR database of the NCBI, we carried out BLASTX with the software DIAMOND v.0.9.21 [37] with an e-value cut-off of 1.0e-5. For functional annotation on the GO database, we used BLASTX with DIAMOND v.0.9.21 and an e-value cut-off of 1.0e-5. The GO terms of the main categories: biological process (BP), cellular component (CC), and molecular function (MF) were listed. For functional annotation on the KEGG database, we used BLASTX with DIAMOND v.0.9.21 and an e-value cut-off of 1.0e-5. A bi-directional best-hit method was performed to search against the KEGG database to obtain the KO (reference pathway) number of the KEGG annotation.

To compare the effect of the soil type in the expression of the genes of the habanero pepper transcriptome, a differential gene expression (DGE) analysis was performed using the program DESeq2 [38] in R v.4.2.0. The comparison was established as RS vs. BS, using two biological replicates for each soil type (RS1 and RS2 vs. BS1 and BS2), using as input data the read counts previously determined in the section of abundance estimation of the unigenes to obtain the log₂FoldChange (LFC), p-value and adjusted p-value. A cut-off of LFC ≥ 1.5 was established for the highest differentially expressed genes in RS, and a cut-off of LFC ≤ −1.5 was established for the highest differentially expressed genes in BS. A p-value < 0.00835 was established as a statistical significance value. Additionally, a Benjamini-Hochberg (BH) test was performed to compute the false discovery rate (FDR), and an adjusted p-value < 0.1 was established as a confidence limit for statistical significance [39]. A volcano diagram was used to represent the expression of the genes with LFC values in the x-axe and the −log10 p-value in the y-axe using the ggplot2 package [40] of R. Afterwards, a gene set expression analysis (GSEA) was performed in the transcriptome to obtain enrichment metabolic pathways which were related with the development, quality and the capsaicinoid content in the fruits by the effect of the soil type. First, the read counts of all the unigenes annotated against the same KEGG term were summed for each sample (RS1, RS2, BS1, and BS2), and the sums were used as input data to perform a DGE analysis with DESeq2 comparing the conditions RS vs. BS using the two biological replicates (RS1 and RS2 vs. BS1 and BS2). Then, the LFC values were used to perform the GSEA with the software ClusterProfiler v.4.4.4 [41] with the R packages ggplot2, enrich plot [42] and pathview [43]. GSEA was run with the following parameters: 1000 permutations, a minimum and maximum gene set the size of 3 and 800, respectively, a p-value cut-off of 0.5, without adjusted p-value method, and searching against the KEGG pathway database of C. annuum to obtain a dot-plot of enriched terms.

To compare the difference in the individual expression of the genes of the enriched metabolic pathways and the genes of the capsaicinoid biosynthetic pathway by the effect of the soil type, representative genes of the different pathways were selected for DGE analysis between the soil types. Two representative regulated genes of each enriched pathway were selected in the C. annuum KEGG database after the GSEA analysis. The coding sequences (CDS) of the genes were used as query sequences to search homologs against the unigenes library of the transcriptome by local BLAST analysis using the program Bioedit and selecting the transcripts sequences with an e-value cut-off of 0.0. As representative sequences in the capsaicinoid pathway, genes of key steps of the pathway were selected according to the literature to analyze if the soil type affected its expression [15,24,26,44–47]. For AT3-D1 and AT3-D2 [44] the CDS of the respective sequences (PHU26684.1 and PHU26685.1) in the genome of C. chinense were selected. For pAMT, the CDS of the mRNA of the gene in C. chinense (AF085149.1) was selected [45]. For the ketoacyl-ACP synthase (KAS) and acyl-ACP thioesterase (FAT) genes, the mRNA of C. chinense (AF085148.1 and AF318288.1, respectively) were selected [46]. For the ketoacyl-ACP reductase (KR) gene, the mRNA of C. annuum (EU616561.1) [47] was selected. Also, a MYB transcription factor gene that regulates the expression of genes of the biosynthetic capsaicinoid pathway was selected [46], using the CDS of the mRNA of C. annuum and C. chinense MYB homologues (MK074771.1 and MF062086.1) [25,46]. The selected sequences were used as a query in a local BLAST analysis against the unigenes library using Bioedit to search homolog transcripts with an e-value cut-off of 0.0. A DGE analysis was performed with the software DESeq2 in R v.4.2.0 comparing the conditions RS vs. BS for each gene and using the read counts obtained in the abundance estimation section as input data. Two biological replicates were used for each soil type (RS1 and RS2 vs. BS1 and BS2). A BH test was performed to calculate the FDR and an adjusted p-value < 0.05 or adjusted p-value < 0.1 were considered limit values for statistical significance. A heat map was constructed (in TPM units) to represent the expression of the transcripts of the genes between BS and RS, using the gplots package [48] and RColorBrewer [49] in R v.4.2.0. The TPM values were calculated by the software RSEM v.1.3.1 with the read count values [36].

The expression of AT3-D1 and AT3-D2 genes was determined in habanero pepper fruits at 25, 45, and 60 DAA from plants growing in BS and RS. First, total RNA isolation from fruits (set of five fruits per sample), previously grounded with a blender in liquid nitrogen, was performed by using the Plant/Fungi Total RNA Purification kit (NORGEN BIOTEK CORP, Canada) according to the manufacturer’s instructions. Then, ≈3 μg of total RNA was treated with DNase using the Turbo DNA-free kit (Invitrogen, USA). cDNA synthesis was carried out in ≈1.5 μg of RNA using the SuperScript III Reverse Transcriptase (Invitrogen, USA) following the manufacturer’s instructions.

PCR was made using specific primers for AT3-D1 and AT3-D2 designed by aligning the CDS of AT3-D1 (PHU26684.1) and AT3-D2 (PHU26685.1). Specific primers designed were F-AT3D1 5′-ATCTCTTGCAGAGAGCATAGTTTTG-3′ and R-AT3D1 5′-TGGAACTAAAGTTGTTGTCGTATG-3′ for AT3-D1. For AT3-D2 the specific primers were F-AT3D2 5′-GCTCATGCAGAGGGCATAGTTTTT-3′ and R-AT3D2 5′-AGGAATTAGAGTCGTTGTCATCAT-3′. β-tubulin gene of habanero pepper (PHU20966.1) was used as a constitutive control using the specific primers F-Tub 5′-GGAGGGTGAGTGAGCAGTTC-3′ and R-Tub 5′-CTGCTGTCGCATCCTGGTAT-3′. The PCR reaction was made in 12 μL reactions mix containing 0.2 μL of Taq polymerase (Invitrogen, USA), 1 μL (500 ng) of cDNA, and 0.36 μL (9 pmol) of forward and reverse primers. Gene amplifications were done by an initial denaturalization at 95°C at 30 s; 35 cycles of 30 s at 95°C, 30 s at 53.4°C–55°C and 15 s at 72°C; and a final extension of 72°C at 5 min. Aligning temperature was 53.4°C for AT3-D1, 55°C for AT3-D2, and 54.2°C for the β-tubulin control. Finally, the relative expression of each gene was determined by the comparative method between the target gene and the β-tubulin gene as a reference. Gel-analyzer Software (http://www.gelanalyzer.com/) was used for quantification.

Habanero pepper plants (C. chinense ‘Jaguar’) were cultivated in two soil types: BS and RS, in a cycle of 150 DPT. In a cycle of flowering of 60 d, the number of flowers and fruit production were measured, comparing both soil types (Fig. 1). There was no difference in the flower production, nor the fruit production between soils. The flower production was 305 and 284 flowers in BS and RS, respectively, and the total fruit production was 40 and 35 fruits per plant, respectively. The flower abortion was 85% for the plants in both soil types, so the fruit production was low. This is expected in habanero pepper cultivated in Yucatan, where the environmental temperature is nearly of 40°C in summer, even when C. chinense is tolerant to water deficit [50]. The total fruit production of 45 DAA was not higher than 35 fruits per plant in none of the soils. If the total production is considered, there was no difference in the production parameters between soils (Fig. 1A).

Also, to compare the fruit development, the fruits of 45 DAA, which are starting the maturation stage, were divided into categories according to the fruit large. The production of fruits, total weight, and average weight of the fruits per plant were compared in each rank between soils. The highest number of fruits in this stage is 3–3.99 cm in size, with 19 and 16 fruits per plant for BS and RS, respectively, without differences between soils. The total number of the biggest fruits (≥4 cm in large) was lower (3 and 7 for BS and RS, respectively), however, in RS there was a greater number of biggest fruits than in BS. Also, there was a greater number of fruits between 2–2.99 cm in BS than in RS. There was a similar number of smaller fruits (<2 cm) between soils (Fig. 1B).

When the total weight of the fruits was analyzed (Fig. 1C), it was observed that the fruits between 3–3.99 cm were the heaviest, with 94 g and 73 g of total weight in BS and RS, respectively, and with the fruits in BS with a greater total weight than the fruits of RS. The fruits in the range of 2–2.99 cm also have a greater total weight in BS than in RS, with 36 and 22 g of total weight, respectively. On the contrary, the bigger fruits (≥4 cm) have a greater total weight per plant in RS (51 g) than in BS (26 g). The total weight of the smaller fruits was similar between soils. The tendencies observed in the total weight were like those observed in the total number of fruits (Fig. 1B), which is an expected behavior.

Finally, the average weight of the fruits per plant was analyzed (Fig. 1D). The tendency was an increment in the average weight with the increment in the size of the fruits. The weight was similar in all the ranks, except in the fruits of 3–3.99 cm, in which there was a greater average weight in the fruits of BS (5.30 g) in comparison to RS (5.05 g).

Having all the results into account, for the fruits of 45 DAA, RS is better to harvest a greater number of fruits with bigger size (≥4 cm), and in BS there is a tendency to obtain a major number of smaller fruits (2–2.99 cm). However, both soils have a greater number of fruits of an intermediate size (3–3.99 cm). Also, having into account that the flower abortion, the total flower and fruit production, and the average weights were similar, we conclude that any of both soils is good for cultivating the habanero pepper and harvesting the fruits, even though there is a tendency to obtain bigger fruits in RS. However, this is a little different if we consider that this is part of the fruit development and that the total production is more relevant.

One of the most important parameters to evaluate the quality of the habanero pepper is the determination of the capsaicinoid content. These compounds add value to the fruits, especially the capsaicin and the dihydrocapsaicin, which represent 90% of the capsaicinoids and give the spicy sensation to the peppers, which is an attribute highly appreciated by consumers and’s a fundamental parameter in the pepper quality [4].

The content of capsaicinoids in the fruits of habanero pepper plants growth in BS and RS were compared between soils and in different development stages (25, 45, and 60 DAA) (Fig. 2). In general, the capsaicinoid content was between 7 to 12 mg of capsaicin/g dry weight of fruit (DW). The analysis shows the evolution in the capsaicinoid content between the two soils during the fruit development and determines if there was a difference in the evolution of the capsaicinoid content and in which soil there was higher capsaicinoid content due to the soil used since in the anterior section we observed that the general production parameters were similar between BS and RS. The difference in the content of capsaicinoid in the peppers by the effect of the type of soil was dependent on the development stage analyzed. At 25 DAA there was a difference in the capsaicinoid content, where BS has a higher content (10.72 mg/g DW), concerning RS (8.92 mg/g DW). At 45 DAA the capsaicinoid content reaches the maximum in both soil types. It was similar between soils, but there was a higher and significant increment in the capsaicinoids in RS (11.48 mg/g) concerning 25 DAA. On the contrary, in BS there was no difference in the capsaicinoid content (11.30 mg/g DW) concerning 25 DAA. At 60 DAA there was a higher capsaicinoid content in RS (9.87 mg/g DW) concerning BS (7.72 mg/g DW), but in both soils, the capsaicinoid content decreased due to the capsaicinoid catabolism in advanced development stages.

When we analyzed and compared the evolution in the content of capsaicinoids between soils, the results suggested a difference in the metabolism process during fruit development. In BS, at 25 DAA a maximum was reached without a significant difference to the content at 45 DAA. On the contrary, in RS there was an increment between 25 and 45 DAA, suggesting that the synthesis of capsaicinoids continued predominant in RS until reached 45 DAA. Also, we noted an apparent maximum limit of capsaicinoids at 45 DAA in both soils, and this maximum was reached first in BS at 25 DAA, suggesting that the synthesis was more active in the early development stages in BS. Finally, at 60 DAA, although in both soils there was a decrement in capsaicinoids due to the catabolism predominance, this was smaller in RS concerning 45 DAA, which corresponds to 85.97% concerning the maximum capsaicinoid content at 45 DAA. In comparison in BS at 60 DAA there was a higher decrement respect to the content found at 45 DAA, which corresponds to 68.31% (Fig. 2).

We conclude that if we collect fruits at 45 DAA, the type of soil is not important if we want the capsaicinoid content for industrial applications, and the fruits of 45 DAA can be harvested without waiting for advanced maturation. On the contrary, if we want the complete mature fresh fruit of 60 DAA for consumption or use of the capsaicinoid content in industrial applications, then is better to harvest the fruits in RS, that have a higher capsaicinoid content in complete mature stages.

According to the aim of this study, we wanted to compare the expression of genes of the capsaicinoid biosynthesis pathway between soils. For this purpose, we selected the fruits of 25 DAA for transcriptomic analysis due to it was the difference in the capsaicinoid content between soils in which BS reached a maximum content while in RS the accumulation continued beyond 25 DAA (Fig. 2), suggesting a difference in the level of synthesis metabolism that could be analyzed at transcriptomic level. Also, in intermediate development stages, it is known that the capsaicinoid synthesis is very active in the pepper fruits and the gene expression of the biosynthetic pathway is at high levels [51], so the fruits of 25 DAA were appropriate to compare levels of expression of these genes in the different soils.

For each soil type (RS and BS), two cDNA libraries were generated (biological replicates) having in total four cDNA libraries in paired ends (RS1, RS2, BS1, and BS2). After filtering the raw reads, there were 42.6 a 55.9 million total reads per library. Between 95%–96% of the reads had a base quality of Q30 or higher, and the average quality of the reads was Q36 (Table 1), indicating high quality in the sequences in the four libraries. The reads have 36–101 nt in length, with most of the reads having 101 nt. The size of the libraries was between 4289 to 5625 Mb. The GC composition was similar between samples, which was expected due to the fruits being of the same development stage and pepper variety (Table 1).

After de novo assembling and filtering redundant sequences, a transcriptome of 99159 unigenes was obtained with an average length of 689.55 bp, a composition of 38.95% GC, and 68.37 Mb. The minimum contig length was 201 bp and the maximum was 9421 bp. The N50 was 1143 bp (Table 2A). Similar de novo assembling values have been obtained in transcriptomes of flowers, immature fruits, intermediate mature fruits, and mature fruits in spicy varieties of C. chinense and C. frutescens [52,53]. In the realignment of the reads to the assembled transcriptome, 78%–82% of the reads were mapped in the library of unigenes (Table 2B), which indicates that a lot of the information of the reads is conserved in the assembled transcriptome [54], and allows to use it to search genes of interest, estimate the abundance of the genes and to analyze its levels of expression in the four libraries [32,51,52].

The annotation of the four libraries against the GO database, in the main categories BP, CC, and MF, was made to group the genes in the most abundant GO subcategories (Fig. 3). A total of 30993 unigenes of the transcriptome were annotated against the GO database. The four libraries had a similar proportion of the dominant subcategories. In the BP category, the most abundant subcategories were “cellular process” (15910 unigenes), “metabolic process” (13790), “biological regulation” (7048 unigenes) and “response to stimulus” (5345 unigenes). In CC the most abundant subcategories were “cell part” (24747 unigenes), “organelle” (15865 unigenes), “membrane” (6602 unigenes) and “organelle part” (6304 unigenes). In MF, “catalytic activity” and “binding” (14398 and 12176 unigenes) were the subcategories with more proportion of genes, followed by “transcription regulator activity” (1892 unigenes) and “transporter activity” (1878 unigenes) (Fig. 3).

To compare the effect of the soil type in the transcriptome of habanero pepper fruits, a DGE analysis was performed comparing the conditions RS vs. BS. We used the trimmed reads libraries having two biological replicates for soil type (RS1 and RS2 vs. BS1 and BS2) and the assembled unigenes library estimated the abundance of the transcripts in each sample (as reads counts), and then the DGE was performed in the software DESeq2. Of the 99159 unigenes, 6292 had 0 counts in the four libraries and were discarded for further analysis. The other 92867 unigenes were analyzed with a volcano diagram (Fig. 4A), where the comparison of the expression was in function of the LFC value, which shows how much a gene is expressed in a soil concerning the other, and the p-value, which express the statistical significance of the comparison [55]. Due to the comparison performed as RS against BS, the positive values of LFC represent the genes more expressed in fruits of RS, and the negative values represent the genes more expressed in fruits of BS. A total of 533 unigenes had a differential expression between soils, 83 unigenes had higher expression in BS (LFC ≤ −1.5), and 450 unigenes had higher expression in RS (LFC ≥ 1.5) (Fig. 4A). Between the genes with higher expression in RS, there were F-box proteins, extensins, proline-rich proteoglycan, polyubiquitin-like protein, BURP domain-like protein, leucine-rich repeat receptor-like protein kinase, protein mago nashi homolog, calmodulin-lysine N-methyltransferase and 3-ketoacyl-CoA-synthase (Fig. 4A and Table S2a).

Between the genes with higher expression in BS, there were orthologues of the RuBisCO, chlorophyll-binding proteins, BREVIS RADIX-like, hypothetical proteins of the mitochondrion, albumin-like protein, and RING-H2 finger-like protein (Fig. 4A and Table S2b). We noted a particularly high expression of genes related to photosynthesis and antenna proteins in BS, that was not found in RS, and we concluded that BS is better for activating the expression of the photosynthetic system, which could be favorable for the quality of the fruits.

Posteriorly, the genes were analyzed in sets in function of the metabolic pathways to find metabolic pathways that were influenced by the type of soil. This gene-set analysis has the advantage of analyzing the behavior of the genes of large databases grouping them in categories or sets that allow to analyze tendencies within these categories and that may have a biological significance [56]. In this case, we were interested in finding in which soil there were more represented (enriched), metabolic pathways related to the fruit development, quality, and its capsaicinoid content, which would suggest which soil can be favorable for those metabolic pathways in a transcriptomic level.

The enrichment analysis of metabolic pathways was made taking into account the annotation of the unigenes by its KEGG terms. Of the 99159 unigenes, 42936 were annotated within the KEGG database. A GSEA analysis was performed with a previous DGE analysis, in which all the reads of the contigs annotated within the same KEGG term were summed in each library (RS1, RS2, BS1, and BS2) and comparing the condition RS vs. BS, which allowed us to find metabolic pathways overrepresented in a soil type by the LFC values [41,57]. A total of 11 enriched metabolic pathways were found, with a high statistical significance (p-value) (Fig. 4B). The enriched pathways were photosynthesis-antenna proteins; pentose phosphate pathway; circadian rhythm; anthocyanin biosynthesis; alanine, aspartate, and glutamate metabolism; arginine biosynthesis; proteasome; terpenoid backbone biosynthesis; 2-oxocarboxylic acid metabolism; carotenoid biosynthesis and porphyrin metabolism. The metabolic pathway with the higher number of genes was the pentose phosphate pathway, and the anthocyanin biosynthesis had the lesser number of genes. The pathway with the higher proportion of genes concerning the total number of genes annotated in the pathway (GeneRatio) was the group of the photosynthesis-antenna proteins, and the pathway with minor GeneRatio was the porphyrin metabolism. Other pathways, the pentose and glucuronate interconversions, and monobactam biosynthesis, had low statistical significance and were discarded for further analysis (Fig. 4B).

In the case of RS, the enrichment was found only in the anthocyanin biosynthesis pathway. In BS, most of the pathways were overrepresented. These pathways were related to the photosynthesis (antenna proteins, porphyrin metabolism), sugar metabolism and reduction equivalents (pentose phosphate pathway), biosynthesis and metabolism of secondary metabolites, nitrogen and amino acids (anthocyanins, carotenoids, terpenoids backbone, arginine, alanine, aspartate, glutamate, 2-oxocarboxylic acid) and cellular function related with the protein degradation (proteasome) and circadian rhythm (Fig. 4B). Concerning the capsaicinoid metabolism, the 2-oxocarboxylic acid metabolism is related to the capsaicinoid pathway via the synthesis of the amino acids precursors valine and leucine [15]. However, we did not find an enriched pathway directly related to the capsaicinoids, like the phenylpropanoids pathway or the synthesis of branched-chain fatty acids. This suggests that these pathways were not overrepresented by a particular soil.

The results obtained from the general DGE analysis and the enriched pathways analysis led us to conclude that BS is particularly favorable for a high expression of the photosynthesis genes and the antenna proteins. Also, an important number of metabolic pathways related to primary and secondary metabolism are overrepresented in BS, which is favorable for the quality and the nutrient content of the fruits of habanero pepper plants growth in this soil, at least, in the transcriptomic level at 25 DAA. Taking into account that the global production of fruits was similar in both soils (Fig. 1A), BS would be better for the habanero pepper fruits to stimulate its photosynthetic system and metabolism (porphyrin, sugars, amino acids, terpenoids, carotenoids) which could lead to better nutritional content.

According to the aim of the study, we wanted to compare the expression of genes related to fruit development and the capsaicinoid synthesis between the two soil types. The enriched metabolic pathway analysis shows us that in BS various metabolic pathways related to fruit development were overrepresented, but this only allowed us to analyze the effect of the soil in the pathways as a whole, but not the individual expression of the genes of these pathways. This is important, particularly for regulated or key genes. Also, we did not find an overrepresentation of a pathway related to capsaicinoid metabolism. Therefore, we proceeded to perform a DGE analysis of highly regulated genes of the capsaicinoid metabolism and of genes of the enriched pathways comparing the condition RS vs. BS in a similar way as we did for the global expression analysis.

In this work, we reported the effect of two contrasting soil types (BS and RS) of Yucatan on the production of fruits and the capsaicinoid content in habanero pepper. The results of production showed a floral abortion relatively high (85%) in both soils. The total production was, on average, 35 to 40 total fruits and 31 to 33 fruits of 45 DAA, which are fruits that are nearly in the complete maturity stage. Ramírez-Meraz et al. [60] reported that, in general in open sky and protected agriculture systems, that the habanero pepper ‘Jaguar’ had fruits of 6.5 to 10 g and a size of 3.8 to 5.5 cm in large. In the present work, the bigger fruits (≥4 cm) had similar values (6.95 to 6.96 g of average weight).

When the fruits were compared in ranks of size, in RS there were more bigger fruits (≥4 cm), and in BS there were more fruits with a minor size (2–2.99 cm). This result was similar to the obtained by Medina-Lara et al. [8] in habanero pepper ‘Jaguar’ in RS and BS irrigated with water with high conductivity, and obtained a more contrasting result in the differences in the size of the fruits by soil type, with a tendency in BS to obtain smallest fruits (<1 cm). In the present work, we used water of low conductivity. Compared with Medina-Lara et al. [8], we also obtained in RS bigger fruits, but in BS there is still an important number of developed fruits (3–3.99 cm of large), and the global difference in production was considered small between soils.

BS tends to have a higher content of nutrients than RS [10,11]. But it is important to consider the mobility of the nutrients in the soil matrix to the solution of the soil, and the roots of the plants [61]. The more efficient use of the nutrients of the soil has been observed in RS in comparison to the BS despite the lower nutritional content of the former [8], but these observations have been made in habanero pepper irrigated with water with high conductivity, in contrast to the water used in the present work, so the efficiency in the use of the nutrients in habanero pepper in water of low conductivity should be also studied.

The differences in soil texture (composition of sand, silt, and clay) and content of organic matter are also important for the mobility and efficient use of nutrients [8,62]. These differences in composition influence the structure, porosity, and apparent density of the soil, which affects the water filtration and the retention of the components of the soil matrix [63]. RS tends to have higher apparent density and lower porosity [62]. BS tends to have higher organic matter, which should give them a higher proportion of largest pores that filter the soil solution more than in RS, and this affects the diffusion of the nutrients to the plant roots, but also the organic matter helps to the soil structure through stabilization of aggregation, given cohesiveness to the soil particles. On the other hand, RS has more particles of smaller size with a higher superficial area that could retain the soil components, affecting mobility [63]. Also, the chemical form of the nutrients in the soil must be considered. In BS, there has been reported higher content of nitrate than in RS [10], which is a chemical form of the nitrogen that the habanero pepper can assimilate. Other nutrients, like phosphorus, are in low solubility and availability for the plant [64] if we attend to the pH of BS and RS, which are neutral or have little alkalinity [8,11]. Also, the differences in the microbial populations between RS and BS must be considered, which influenced the chemical form of the nutrients and the organic matter of the soils [11]. For all these reasons, the chemical composition of the nutrients and the organic matter must be investigated in more depth to have a better understanding of the availability of the components of the soil for the habanero pepper.

The capsaicinoids are considered the most important metabolites in the habanero pepper quality for the spicy sensation that they give to consumers and the bioactive properties that are of industrial interest [13]. For this, is important to determine the effect of the soil type over the capsaicinoid content in the pepper fruits to know which soil is more convenient if we want to produce fruits with higher capsaicinoid content. The results showed an increment in the capsaicinoid content between 25 and 45 DAA in RS, and later a decrease at 60 DAA. In BS the maximum capsaicinoid content was reached at 25 DAA and only decreased at 60 DAA. The higher capsaicinoid content was between 11.30 to 11.48 mg/g DW or 169500 to 172200 Scoville Heat Units (SHU) in BS and RS, respectively. This pungency was superior to those showed in other domestic species of Capsicum (C. annuum, C. frutescens, C. pubescens, and C. baccatum) [46], but lower than other varieties of habanero pepper of Yucatan [65]. Also, the capsaicinoid content was similar to the obtained by Medina-Lara et al. [7] for habanero pepper in RS, in which, under urea fertilization, there was not possible to increment the capsaicinoid content after reaching a limit, without difference with the treatment without fertilization [7]. It has been reported that is possible to obtain higher capsaicinoid content under organic fertilization in soils with contrast characteristics, as has been reported by Das et al. [6] in C. chinense ‘Borbhut’ in two different soil types in India, with different characteristics in texture, nutrients and organic matter, similar to the contrast between BS and RS. Comparing the results of the present work, the two soils are good to obtain a similar capsaicinoid content at 45 DAA, which are fruits reaching the maturity stage, without the necessity to apply specific nitrogen fertilization, but considering the results at 60 DAA is more convenient to harvest the fruits in RS to obtain mature fruits with higher capsaicinoid content.

Concerning the evolution in the capsaicinoid content during fruit development, the results suggest an effect in the synthesis, accumulation, and degradation of the capsaicinoids, with different effects in each development stage. Considering that the capsaicinoid content in a determinate point is the reflection of its synthesis and degradation, then, between 25 and 45 DAA in RS predominate the synthesis, increasing the capsaicinoid content. In BS, neither the synthesis, nor the degradation seems to predominate, and the synthesis should occur in early stages, so between 25 and 45 DAA the global capsaicinoid metabolism did not change the capsaicinoid content. At 60 DAA, in both soils, the catabolism of capsaicinoids seems to predominate, so the content of these metabolites decreased, and the effect was higher in BS. This indicates that the soil type affects the metabolism of capsaicinoids. The expression of genes like PUN1 is sensitive to the type of soil [27]. One possibility is that the nitrogen content, its availability [8], or the chemical form in which it is presented in the soils affects the gene expression of the capsaicinoid metabolism of its enzymes [23]. It is known that the nitrate content in the fruits and the placenta of the habanero pepper affects capsaicinoid accumulation [66]. The differences in the organic matter and the humic and fulvic acids between soils also is related to the capsaicinoid content [6], and the differences in organic matter between RS and BS probably are another factor that affects the changes in the capsaicinoid accumulation [8]. According to the results of the present work, in BS the capsaicinoid synthesis probably has been stimulated more strongly in the early stages, reaching a maximum at 25 DAA, before the maximum in RS. So, we selected this development stage for the transcriptomic analysis and to find if the gene expression of the capsaicinoid metabolism was related to the differences in the capsaicinoid accumulation.

We analyzed the fruits of 25 DAA by transcriptomic to compare the expression of genes related to capsaicinoid accumulation. These intermediate development stages are known to be active in the upregulation of the genes related to capsaicinoid biosynthesis [51,56].

The genes in each library (BS1, BS2, RS1, and RS2) were annotated against the GO database. This allowed us to group them into the three main functional categories: BP, CC, and MF [67]. The proportion of the genes in the subcategories in each one of the three main terms was similar. This suggests a similar functionality in the habanero pepper fruits of 25 DAA, independent of the type of soil, in the main GO categories.

When the global expression of the genes was analyzed and compared between soils, in BS many of the higher expressed genes (LFC ≤ −1.5) were related to the photosynthesis. There were orthologues of chlorophyll binding proteins of the LHC, and RUBisCO. It is known that the expression of genes related to photosynthesis is controlled by the development stage of the fruits, decreasing its expression in fruits in ripening stages with the change in coloration [68,69]. In this work, the fruits of 25 DAA did not show the difference in the coloration between soils, but at 45 DAA, when the fruits were in ripening or nearly in the complete mature stage, in BS there was a higher proportion of fruits with a phenotype of change in coloration and less proportion of green fruits in comparison to RS (results not show). It is possible that in BS the photosynthetic system was more stimulated in fruits of 25 DAA than in RS, and then in advanced development stages the maturation is favored for obtaining fruits with higher sugar and secondary metabolites content [70,71]. Other orthologues with higher expression in BS were RING-H2 finger protein, which is related to the growth, development, and response to abiotic stress [72] and with seed development [73]; BREVIS RADIX, related to the development of the fruits and the number of locules in C. annuum [74]; albumin-like, that has been found in seeds of C. annuum, serves as storage and had antimicrobial properties [75].

In RS, between the higher expressed genes (LFC ≥ 1.5) there was a 3-ketoacyl-CoA synthase, related to the fatty-acid synthesis. Another gene was an orthologue of the calmodulin-lysine N-methyltransferase, where the calmodulin is related to the calcium signaling between the cell wall and the cytoplasm, and in the growth of the pollen tube [76,77]. Other genes with higher expression in RS were orthologues of the mago nashi protein, related to the fertility [78], proline-rich proteoglycan; extensins; F-box proteins; polyubiquitin; and BURP domain protein, which participate in different processes like the response to environmental stimuli, floral development and in the fruit and its cellular wall development, including in the stages of maturation, softening and abscission [79–82].

The main difference in the higher expressed genes between BS and RS was that, although in both soils there were genes related to the general fruit development, RS was more specific in the genes related to growth and development, including the flowering and maturation of the fruits, the metabolism of the cell wall and the softening and the abscission of the fruits, which could indicate that RS is favorable in the growth of the fruits and in the softening of the fruits. This could have influenced that there were bigger fruits of 45 DAA (≥4 cm) in RS than in BS (Fig. 1B). In BS, there was specifically higher expression of genes related to the photosynthesis and the chlorophyll, like RUBisCO and chlorophyll-binding proteins. We suggest that this is related to the higher organic matter and higher content of nutrients in BS which can stimulate the metabolism of habanero pepper fruit and the biosynthetic processes for primary and secondary metabolites [6] more than RS, a soil with less organic matter and nutrients [6,11].

The enrichment analysis of metabolic pathways showed that most of the overrepresented pathways were in BS. Between there, there were essential metabolic functions like photosynthesis, where we found photosynthesis-antenna proteins and the porphyrin metabolism; the pentose phosphate pathway (that generates NADPH and pentoses for the biosynthetic processes); the proteasome; the circadian rhythm; and the nitrogen and amino acids metabolism. The inclusion of overrepresented photosynthetic-related pathways together with the presence of higher expressed genes related to photosynthesis and also the other overrepresented pathways related to primary metabolic processes, suggest that BS is better for the metabolic and physiologic development of the fruits. Also, other overrepresented metabolic pathways were related to the synthesis of metabolites of interest (terpenoids, carotenoids, and amino acids). If we take into account that these metabolites have properties that determine the fruit quality and are beneficial for consumer health [3], the results suggest that BS stimulates the expression of genes of the metabolism of the fruits better than RS, which could lead to obtain better quality in the fruits, enhancing biosynthetic processes and more content in nutrients like sugars, amino acids, terpenoids, carotenoids, and capsaicinoids, which also has been observed in other works that use soils rich in carbon and organic matter, and nutrients like nitrogen and phosphorus [6]. Related to the capsaicinoid biosynthetic pathway, some enriched pathways were indirectly related to the capsaicinoids, like the glutamate metabolism [23], the 2-oxocaborxilic acid metabolism, which is related to the synthesis of the branched-chain amino acids leucine and valine, precursors of the capsaicinoids [83], or the synthesis of anthocyanins, that are part of the phenylpropanoids pathway [84]. In general, these pathways seem to be more related to the fruit development than to the relative accumulation of capsaicinoids at 25 DAA.

In line with the study objective, our aim was to compare the precise gene expression patterns associated with development and capsaicinoid synthesis. To achieve this, we carefully selected a comprehensive set of thirty-one genes from enriched pathways and biosynthetic pathways of capsaicinoid production. Subsequently, we conducted a differential gene expression (DGE) analysis for these genes. Additionally, we examined the expression of two copies of AT3, a pivotal gene in the capsaicinoid biosynthesis pathway, across various developmental stages to compare its behavior in different soil conditions.