Background & Summary

The Moroccan fir (Abies marocana Trab.), is a coniferous tree from the Pinaceae family, endemic to a secluded area of approximately 4,770 hectares in the western Rif Mountains of northern Morocco1. Ecologically and economically, A. marocana forests offer different resources, like wood, while supporting biodiversity, climate regulation, and soil protection2. The species is classified as endangered and included on the IUCN Red List, with projections pointing to a decrease in its potential ecological range under climate change scenarios, posing serious threats to its conservation1.

The genetic characteristics of the Moroccan fir have been considerably studied, through microsatellites3,4,5 and single nucleotide polymorphisms6,7,8. However, to our knowledge, no transcriptomic data was available for the species to date. Hence, the main objective of this study was to develop a robust transcriptome dataset for the Moroccan fir to advance genetic research on the species. Here, we present a highly complete de novo transcriptome assembly for A. marocana. To maximize gene expression and obtain an extensively comprehensive Moroccan fir transcriptome, samples were also exposed to different experimental environmental conditions, such as drought, heat, and cold, and transcripts were assembled using RNAseq data from three different organs (leaf, stem, and root) generated through both long- and short-read sequencing technologies.

Overall, this work ultimately aims to support conservation initiatives to protect A. marocana populations amid ongoing climate challenges. For instance, it could help uncover the molecular mechanisms underlying the species’ response to environmental stress; an area explored through gene expression experiments in other conifers9,10,11,12, which holds relevant importance given the growing threats that climate change poses to global forest ecosystems13.

Methods

Samples and applied environmental conditions

The field sampling was carried out in the Talassemtane National Park (see14 and references therein). We investigated the two main populations of A. marocana, located in the Tazaot range (35.25 N, −5.09 W, around 1760 m a.s.l.) and the Talassemtane range (35.14 N, −5.14 W, around 1730 m a.s.l.). Mature cones were sampled in late October from randomly selected trees, at least 100 m apart, in both populations, using a telescopic pruner or by climbing the trees (see details in14). Seeds were grown during three years in a greenhouse at the Pablo de Olavide University (Sevilla, Spain), with a photoperiod of 12 hours of light/12 hours of darkness (light at 600 lux from 7 AM to 7 PM) and watering of 10 ml/day, using 0.77 cm3 pots with sterile sand and perlite substrate at a 3:1 ratio15.

A total of 21 Abies marocana saplings were subjected to seven different environmental conditions (Table 1) (that is, 3 biological replicates per condition, which were then pooled before sequencing) in control growth chambers (CGC) at the Pablo de Olavide University. Leaf, stem, and root samples were collected from each sapling at the conclusion of its respective exposure to the different environmental conditions. The experimental conditions were chosen to capture a wide spectrum of ecologically meaningful stress factors that are known to provoke distinct transcriptional responses in forest tree species, including thermal extremes16,17,18, drought15, hormonal stimulation19, and mechanical damage20. Some of them consisted of brief exposure to severe conditions (for example, −20 °C or 45 °C for 2 h), expected to rapidly activate stress‐responsive gene expression. In contrast, others represented less intense but prolonged challenges (such as 4 °C for 48 h), which likely require extended exposure before measurable transcriptional changes occur. These factors reflect some of the most common environmental pressures experienced by natural populations of silver fir, including simulated herbivore damage through mechanical wounding. The hormonal treatment differed from the others by providing a “positive” physiological cue, as abscisic acid is recognized for its role in promoting growth rather than inducing stress.

Table 1 Details of the seven environmental conditions applied to the samples in this study, including temperature, relative humidity (RH), and other stress factors.
Full size table

RNA extraction and sequencing

Total RNA was extracted from 100 mg of leaf, stem and root of each sample using the Spectrum™ Plant Total RNA Kit (Sigma-Aldrich, USA). RNA quality and quantity were evaluated using RNA ScreenTape (TapeStation 2200, Agilent Technologies), with all samples achieving an RNA integrity number (RIN) > 7. Samples were then pooled by organ, resulting in three pools of 21 samples each, with a minimum of 1000 ng of RNA per pool. Because the main objective of this study was to capture the widest possible transcript diversity for building a comprehensive reference transcriptome and not describing condition-specific differential expression, individual samples were combined before library preparation, and, consequently, no statistical testing or comparisons between conditions were conducted.

mRNA libraries were constructed and sequenced at Macrogen (Korea). Short-read libraries were prepared using the TruSeq Stranded mRNA Library Prep Kit and sequenced on the Illumina NovaSeq X platform (7 Gb per organ pool). Long-read sequencing was performed using PacBio Iso-Seq libraries on the PacBio Sequel II system (9 Gb in total across the three organs) and processed to obtain HiFi reads by Macrogen.

Data pre-processing and de novo transcriptome assembly

Short raw reads were pre-processed using fastp v0.23.421 to remove adapter contamination, poly(A) tails (--trim_poly_x) and low-quality reads (quality < 30 using -q 30; length < 36 using -l 36). HiFi long raw reads were filtered with filtlong v0.2.122, discarding the lowest-quality 10% and reads shorter than 1000 nucleotides. Then, three de novo transcriptomes were assembled using Trinity v2.15.123 with default parameters and the “--long_reads” option, generating separate assemblies for each organ (so while they are technically distinct for each organ, they do not constitute independent biological data, as all reads originate from the same set of pooled individuals). Poorly supported transcripts were removed based on their expression levels, measured in Transcripts Per Million fragments mapped (TPM < 1), using the Trinity scripts align_and_estimate_abundance.pl, abundance_estimates_to_matrix.pl, and filter_low_expr_transcripts.pl. Because reads from all individuals were pooled by organ, transcripts that remain below this threshold are consistently lowly expressed across the dataset, ensuring that only well-supported isoforms are retained while transcriptional noise or assembly artifacts are discarded24. Next, SuperTranscripts were generated and clustered, using the Trinity script Trinity_gene_splice_modeler.py, which collapses unique and shared sequence regions among splicing isoforms into a single linear sequence, representing all exons of a gene without redundancy25. Finally, the three de novo transcriptomes were merged, and a non-redundant assembly was created using CD-HIT-EST v4.8.126, removing transcripts with a global sequence identity of >95%, to collapse highly similar and partially redundant sequences generated by independent assemblies, while maintaining separation between closely related but distinct transcript variants.

The quality of the resulting transcriptome was assessed using transcript number, mean length, total length, and the N50 value, calculated with SeqKit v2.10.027, while transcriptome completeness was evaluated with BUSCO v5.8.028 using the Embryophyta lineage dataset (embryophyta_odb10.2019-11-20, with 1614 conserved land-plant genes).

Functional annotation

Homology-based functional annotation of the transcripts was performed using Blast2GO29 from OmicsBox v3.2.9. DIAMOND BLASTx was run with standard sensitivity against the NR database (2025-03-05), filtered for Viridiplantae, retaining the top five hits. Subsequently, InterProScan was used to annotate proteins with families and domains, following the OmicsBox functional annotation pipeline. Finally, KAAS (KEGG Automatic Annotation Server)30 was used to identify KEGG (Kyoto Encyclopedia of Genes and Genomes) pathways represented by large numbers of annotated transcripts, using the single-directional best hit (SBH) method with 18 plant species as references, including all available tree species and Arabidopsis thaliana due to its highly detailed annotation.

Data Records

Raw reads are available at NCBI’s Sequence Read Archive (SRA), under project SRP570251 (BioProject with accession number PRJNA1235699)31, with BioSample accessions from SAMN47340412 to SAMN47340417 and run accessions from SRR32684349 to SRR32684354. The final assembled transcriptome is available at NCBI’s GenBank Transcriptome Shotgun Assembly (TSA) database under accession GLLU0000000032 and can also be accessed at Zenodo (DOI: https://doi.org/10.5281/zenodo.15023692)33, along with additional annotation files, including GO and KEGG orthologs.

Technical Validation

Transcriptome assembly

A total of 4,074,116 long reads (~7.5 Gbp) and 102,756,401 short reads (~15 Gbp) passed the filtering criteria (Supplementary Tables 1, 2). The de novo assemblies generated approximately 150,000, 200,000, and 250,000 transcripts for the leaf, stem, and root samples, respectively (Table 2). Following expression filtering, the creation of SuperTranscripts, merging of all three assemblies, and clustering by similarity, the final transcriptome assembly comprised 279,439 sequences, with a mean length of 581 bp (ranging from 188 bp to 19,960 bp) (Table 2), and an N50 of 1,405 bp. Of these, 15.9% (44,395) are exclusive to leaves, 30.4% (84,886) are only present in stems, and 34.1% (95,345) are specific to roots, with 19.6% (54,813) shared between the organs (Fig. 1).

Table 2 De novo assembly results for each organ (leaf, stem, root) before and after each step of data processing, including the final version (merged and non-redundant).
Full size table
Fig. 1
Fig. 1
Full size image

Venn diagram showing coincidences between the transcripts expressed in each organ (leaf, stem, and root).

The BUSCO analysis yielded a completeness value of 92.1%, with 1,487 complete BUSCO marker genes, 2.9% fragmented (46 genes), and 5.0% missing (81 genes). Reported BUSCO completeness values for transcriptomes of related conifer species span a broad range, from approximately 50% in Pinus radiata34, 61–83% in Pinus sylvestris35, 80.7% in Pseudotsuga menziesii36, 87.3% in Sequoia sempervirens37, to 94.1% in Picea glauca38 and 94.8% in Abies alba39. The completeness achieved here therefore lies within the upper range currently reported for gymnosperm transcriptomes, supporting the overall quality of the assembly despite the large number of recovered transcripts.

Overall, the presented transcriptome, consisting of 279,439 sequences and a completeness of 92.1%, assembled combining data from both short- and long-read sequencing across different organs and environmental conditions, constituted a valuable resource for future studies on the Moroccan fir.

Functional annotation

Of the 279,439 transcripts, 126,305 (45.2%) were successfully annotated using BLAST or InterProScan. Because the annotation process can associate multiple functional terms with each transcript based on their similarity with previously described sequences, the transcriptome annotation includes a total of 6,744 distinct Gene Ontology (GO) terms and 4,047 KEGG orthologs. The most abundant GO terms (Fig. 2) include “translation”, “RNA modification”, and “protein phosphorylation” in the Biological Process category; “membrane”, “cytoplasm”, and “nucleus” in the Cellular Component category; and “ATP binding”, “protein binding”, and “RNA binding” in the Molecular Function category. In the pathway analysis, metabolic pathways, encompassing carbohydrate, energy, lipid, nucleotide, and amino acid metabolism among others, were notably represented in the assembly (Fig. 3). Genetic information processing pathways, including transcription, translation, and replication, were well-represented too (Fig. 3). Cellular processes, comprising different transport and catabolism and cell growth and death processes, and plant specific pathways involved in environmental information processing (MAPK signaling pathway – plant and Plant hormone signal transduction) and adaptation (Circadian rhythm – plant and Plant-pathogen interaction) were likewise captured in the assembly (Fig. 3). Overall, the dominant GO terms and KEGG pathways reflect core biological processes and highlight that essential processes for tree survival and development have been captured in the assembled transcriptome.

Fig. 2
Fig. 2
Full size image

Summary of the 10 most represented gene ontology (GO) terms from each category (Biological Process, Cellular Component, and Molecular Function) in the assembled transcriptome.

Fig. 3
Fig. 3
Full size image

Summary of the KEGG pathways results, showing the number of different genes in the assembly included in different KEGG categories.

All in all, these annotations serve as a valuable resource for exploring processes, structures, functions, and pathways in A. marocana, offering a high-quality representation of the key molecular processes and biological functions within the species.