Genomic data, sequencing and assembling

A natural population of D. amaguana [43] was collected from Refugio de Vida Silvestre Pasochoa. The Pasochoa Forest is located in the Pichincha province (Ecuador), and the collection area is situated at an altitude of 3,310 meters above sea level (m.a.s.l) (0°26’00” S, 78°29’00” W). Genomic DNA (gDNA) from a pooled sample of 10 individuals (5 males and 5 females) was extracted using the protocol described by Piñol et al. [45]. A NanoDrop™ 2000 ThermoScientific spectrophotometer was used to evaluate the quantity and quality of the extracted gDNA: 8,754 ng/μL and 2.16 A₂₆₀/A₂₈₀ ratios, respectively. The TruSeq DNA PCR library (Illumina Inc., San Diego, CA, USA) was prepared for sequencing on an Illumina NovaSeq 6000 platform with paired-end reads (2 × 150 bp). First, the raw read data were processed, and their quality was assessed using FastQC version 0.12.1 [46]. The FastQC report corresponding to the paired-end raw reads is provided in S1 File. Since the mean quality score across all bases was above Q25 and no adapter sequences were detected, no trimming was performed, allowing the reads to be used as obtained for downstream analyses. Subsequently, a de novo scaffold-level assembly for D. amaguana was performed using Maryland Super Read Cabog Assembler (MaSuRCA) version 4.0.0 (https://github.com/alekseyzimin/masurca) [47] using the previously generated paired-end sequences. Quality metrics and gene prediction were assessed using Python script, assembly_stats version 0.1.4 (https://github.com/MikeTrizna/assembly_stats) [48], and Benchmarking Universal Single-Copy Orthologs v.5.8.3 (BUSCO) [49], respectively. BUSCO was run using the diptera_odb12 lineage dataset (5,067 genes from 76 genomes) and Augustus as gene predictors in the optional parameters. Finally, we used the Redundans pipeline version 2.01 (https://github.com/Gabaldonlab/redundans) [50], with default parameters, to evaluate whether the D. amaguana genome assembly exhibited high redundancy due to duplicated contigs. This analysis was performed without applying any modifications to the original assembly and served as a validation step for its use in all subsequent analyses.

This study was performed using the permit MAE-DNB-CM-2015–0030 assigned by the Ministerio del Ambiente, Agua y Transición Ecológica (MAATE) of Ecuador.

Transposable elements de novo identification

To generate a non-redundant de novo TEs library, we conducted an initial exploration using three different bioinformatic tools to analyze the genome sequence of D. amaguana. The first pipeline used was the Extensive de novo TE Annotator (EDTA) (v.2.1.0; https://github.com/oushujun/EDTA) [51], which was run with default parameters to automatically execute the entire pipeline and compile a final de novo TE library. The D. amaguana genome was analyzed using RepeatModeler (v.2.0.3; http://www.repeatmasker.org/RepeatModeler/) [52] and RepeatMasker (v.4.1.5; https://www.repeatmasker.org/RepeatMasker/) [53] maintaining all default parameters while also adding the -LTRStruct parameter to build another de novo TE library. Finally, the D. amaguana genome was analyzed using the default parameters of reasonaTE (https://github.com/DerKevinRiehl/transposon_annotation_reasonaTE), a transposon annotation tool part of the TransposonUltimate package (https://github.com/DerKevinRiehl/TransposonUltimate) [54]. This analysis provided a FASTA file containing all TE copies identified in the genome. These programs allowed us to perform an initial verification of the TE composition in the D. amaguana genome.

Manual curation of TE libraries and TEs annotation

We manually checked the TE dataset obtained previously using each automatic de novo tool. First, we concatenated the TE library outputs from the EDTA and RepeatModeler into one FASTA file. The concatenated file from the de novo TE libraries generated by EDTA and RepeatModeler was then used for manual curation. We excluded the reasonaTE output because the pipeline did not generate a TE library; instead, it annotated all the TE copies. The curation process was carried out using the Manual Curator Helper tool (MCHelper) (https://github.com/GonzalezLab/MCHelper) [55] in the fully automatic mode. As a result, we obtained an automatically curated TE library. Then, we manually inspected each TE consensus sequence using the Manual Inspection Module from MCHelper, which uses plots generated by TE-Aid (version 1.0; https://github.com/clemgoub/TE-Aid) [56], as well as Multiple Sequence Alignment Plots from CIAlign [57], and other useful structural information such as terminal repeat length and coding domain presence. We inspected each TE consensus sequence and filtered out those that corresponded to false positives (i.e., simple repeats or chimeric sequences). The D. amaguana genome was masked using RepeatMasker v. 4.1.2-p1 [53] with the manually inspected TE library, including TE libraries from other Drosophila species obtained from the Berkeley Drosophila Genome Project (BDGP) dataset (https://www.fruitfly.org/) [58] and the Manual Curated TE library (MCTE) [34] applying the following parameters: -lib -gff -nolow -no_is -norna. We then used the OneCodeToFindThemAll script [59] in the genome annotation to defragment the copies. TEs annotation was analyzed using an in-house Python script (Dynamics_Dama.ipynb) to obtain information on TE copy frequency and size distribution at the superfamily level, as well as TE proportion content and landscape at the order level. Additionally, an in-house Bash script (TEcopies_sequences.sh), utilizing the convert2bed tool from BEDOPS version 2.4.41 [60] and BEDTools version 2.18 [61], was used to extract all TE copy sequences from the whole genome. All these analyses were performed considering the full element length, from start to end position, as defined in the TE annotation file. Both scripts are available at the Zenodo repository (https://doi.org/10.5281/zenodo.16782536). Finally, we ran the Tandem Repeat Finder (TRF) program (https://github.com/Benson-Genomics-Lab/TRF) [62] to identify an estimate of other simple repeats in the masked genome. All the analyses were performed using the manually curated TE library.

TE landscape

To visualize a brief evolutionary TE history and explore the approximate insertion times of TEs –whether TE copies were inserted more recently or correspond to older copies with greater accumulated mutations– at the order level in the D. amaguana genome, we constructed a repeat landscape based on the TE annotation using the in-house Python script (Dynamics_Dama.ipynb). We obtained the divergence percentage directly from the RepeatMasker output, which was calculated relative to the consensus sequence used to annotate each copy. This divergence corresponds to the percentage of nucleotides that do not match the consensus sequence in the manually curated library and each TE copy found in the genome, and was calculated using the following formula:

Clustering analysis to identify satellite DNA (satDNA)

We used Illumina paired-end sequencing raw data from D. amaguana to perform de novo clustering analysis with RepeatExplorer2 [63] on the Galaxy platform [64] for unsupervised identification of satDNAs. Initially, we randomly extracted 20 million paired-end FASTQ files using the Seqtk tool version 1.4 (https://github.com/lh3/seqtk) and uploaded them to the RepeatExplorer Galaxy server (https://repeatexplorer-elixir.cerit-sc.cz/galaxy/). We then checked the read quality using the FastQC tool version 0.12.1 [46], and once again, we randomly subsampled 759,425 reads from each paired-end file. This corresponds to 0.5x of genome coverage. After, the “preprocessing of fastq paired-reads” tool included on the RepeatExplorer Galaxy allowed removing adapters and filtering reads, excluding those with more than 5% of its sequence in low-quality bases. We set the cutoff quality value at Q10 and trimmed the reads to 150 bp. Filtered paired-end files were concatenated into a single FASTA file, which was the input for the RepeatExplorer pipeline. We used the following parameters: REXdb = Metazoa version 3.0, queue = “basic”, advance settings = “yes”, use custom repeat database = “yes”, and by default the threshold for clustering was 90% and assembly was 55%. The results were downloaded for further inspection, considering at first stage only putative satDNA sequences with not less than 0.01% of the genome proportion. From the selected group, a second filter was applied to maintain only those clusters that complied with three out of the four parameters: C-value ≥ 0.8, P-value ≥ 0.7, satellite probability ≥ 0.1, and globular or ring graph layout. Finally, we compared the corresponding consensus of the putative satDNA sequences using BLASTN [65] with known sequences reported from other Drosophila species by De Lima & Ruiz-Ruano [66] and Silva et al. [67], as well as with sequences available in NCBI.

An initial insight of de novo TEs identified in D. amaguana

Three different tools were used for the de novo identification of TEs and to estimate their composition within the D. amaguana genome. The non-curated results from EDTA showed 15.77% total TEs in the entire genome, 7.91% in retrotransposons, and 7.86% in DNA transposons (Fig S1A in S3 File). On the other hand, using the non-curated results from RepeatModeler we obtained a de novo TE library that was masked in the genome with RepeatMasker, here, we identified 19.90% of total TEs. However, 10.71% of the TEs sequences were not classified. The corresponding percentages for retrotransposons and DNA transposons were 4.81% and 4.38%, respectively (Fig S1B in S3 File). Finally, with reasonaTE, we only reported 8.24% of TEs in the D. amaguana genome. This report indicated that 6.27% of the genome corresponded to DNA transposons and 1.97% to retrotransposons (Fig S1C in S3 File).

The initial TE genome composition at the order level in the D. amaguana genome differed, depending on the tool used (Table 2). EDTA annotated that long terminal repeat (LTR) and terminal inverted repeat (TIR) elements occupied 7.41%, and 6.48% of the assembled genome, respectively. Meanwhile, both long interspersed nuclear elements (LINEs) and Helitrons had contents of less than 1.5%. RepeatModeler showed the highest percentage of unclassified sequences and was the only tool to classify TE sequences into short interspersed nuclear elements (SINEs) and Penelope-like elements (PLEs), both comprising less than 1% of each. These elements were reported either by EDTA or reasonaTE, but at almost 0.01% or less. In addition, Helitrons predominated over others, followed by LTR, LINE, and TIR elements. However, the TE composition indicated low percentages (less than 3%) in all orders when using RepeatModeler. Finally, reasonaTE identified only three transposon orders. TIR elements were the most abundant group (6.03%) in the genome, while Helitrons had a proportion of only 0.20%. This value was the lowest when compared to reports using EDTA (1.37%) and RepeatModeler (2.79%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. TE percentage in the D. amaguana genome using non-curated libraries from de novo tools.

https://doi.org/10.1371/journal.pone.0337390.t002

In some cases, other orders for TEs were identified at very low percentages. The summary report from EDTA indicated the presence of mobile elements, such as PLE, Maverick, and unclassified non-LTR sequences, which represented 0.01% of the total genome. Neither RepeatModeler nor reasonaTE reported Maverick in their analysis. However, EDTA did not identify SINEs. This order was identified by RepeatModeler at 0.26% and reasonaTE at a minimal percentage. Moreover, reasonaTE was the only method to classify 0.04% of the genome, specifically as miniature inverted-repeat transposable elements (MITEs) (Table 2).

Overview of the curated TE dataset

Manual curation of TE libraries enabled us to characterize the mobilome in the assembled D. amaguana genome more precisely. Using de novo TE libraries obtained using EDTA and RepeatModeler, we built a curated TE dataset for D. amaguana. This dataset included 737 consensus elements, of which 292 (39.62%) were classified as Retrotransposons (Class I) and 445 (60.38%) as DNA transposons (Class II). Moreover, only 21 consensus sequences complied with the 80-80-80 rule [3], which is similar to the previously described transposon sequences (Table S2 in S2 File). Of these, 20 sequences corresponded to transposons found in other Drosophila species including D. melanogaster, D. hydei, D. mercatorum, D. willistoni, D. ananassae, D. virilis, D. persimilis, D. elegans, and D. bipectinata. The remaining sequence has been previously reported in Lepeophtheirus salmonis (Siphonostomatoida: Caligidae). Therefore, the 716 consensus sequences could be catalogued as potentially novel TE families that have never been described before.

Genome annotation was carried out with this own curated TE dataset, including other Drosophila TE libraries, such as the BDGP dataset [58] and the MCTE library [34]. Subsequently, 942 TEs were annotated in the D. amaguana genome, with approximately 53.18% corresponding to Class II elements and 46.82% to Class I elements (Table 3). Among these, 226 corresponded to known families. Furthermore, we identified 12 defined superfamilies of retrotransposons (LTR: Gypsy, Copia, Bel-Pao, LARD, TRIM; and LINE: CR1, Jockey, I, R1, R2, RTE, Loa) and 11 superfamilies of DNA transposons (TIR: hAT, P, Tc1-Mariner, Pogo, PiggyBac, CMC, MULE-NOF, Transib, Merlin; Helitron: Helitron; and Maverick: Maverick) (see more details in Table 3).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Summary of TE annotation using the manually curated TE dataset for D. amaguana.

https://doi.org/10.1371/journal.pone.0337390.t003

The dataset, including the manually curated transposable element (TE) library with consensus sequences for D. amaguana, genomic TE annotations, and the sequences of all TE copies, is available on Zenodo: https://doi.org/10.5281/zenodo.16782536.

Of the 737 consensus elements, we further classified the TEs based on their potential autonomy. We identified 150 autonomous TEs (Table 4), indicating their potential to be mobilized. In contrast, 587 TEs were identified as fragmented remnants (Table 4), lacking the essential features for autonomous transposition, such as intact ORFs or terminal sequences. This classification was based on the decision tree described in the supplemental materials of the MCHelper tool [55], which is crucial for distinguishing between potentially functional and non-functional transposons, allowing us to better understand the mobilization potential of the D. amaguana genome.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Number of autonomous and non-autonomous transposable element (TE) orders identified in the D. amaguana genome using the manually curated TE library.

https://doi.org/10.1371/journal.pone.0337390.t004

TEs distribution along the entire genome of D. amaguana

We reported that, at the TE order level, Helitrons were the most abundant TEs in the entire genome, representing 30.64% of all TE copies (Fig 1). This corresponded to 99,356 copies and was represented by the only superfamily known to date, Helitrons (Table S3 in S2 File). SINEs and DIRS had the lowest number of copies (657 and 569, respectively) (Table S3 in S2 File). Furthermore, the remaining TE copies were distributed among the four orders (TIRs, LTRs, LINEs, and MITEs). TIRs elements were predominant, with 32,614 copies more than LTRs, 62,263 copies more than LINEs, and approximately 66,659 copies more than MITEs (Table S3 in S2 File). This makes them the second-most abundant TE order in D. amaguana. Among the TIR transposons, the Tc1-Mariner and hAT superfamilies had an important number of copies, representing 18.96% of all TIR copies. However, the largest number of copies of TIRs corresponded to TEs, which could not be classified at the superfamily level within this order, totaling 72,667 copies (Fig 1 and Table S4 in S2 File). The Gypsy superfamily contributed 25,029 copies that almost corresponded to at least half of the copies in the LTR retrotransposons, whereas the CR1, Jockey, and R1 superfamilies similarly dominated the order of LINEs (Table S4 in S2 File). On the other hand, we found that the Loa, RTE, MULE-NOF, Merlin, and Pogo superfamilies had fewer than 100 copies, while the PiggyBac and R2 superfamilies had fewer than 200 copies (Fig 1 and Table S4 in S2 File). In both cases, this represented a genome contribution less than or equal to 0.01%.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Abundance and genomic occupancy of transposable elements in D. amaguana based on annotations from a manually curated TE library.

Each bar represents the total frequency of all copies annotated in the genome of D. amaguana. The names of the order/superfamilies are represented on the y-axis, whereas the frequency of the number of copies is represented on the x-axis. Pie chart showing the genomic occupancy percentages of each transposable element (TE) order in the assembled genome. The pie chart illustrates the relative contribution of Helitrons (6.35%), LTR elements (5.13%), TIR elements (3.63%), LINEs (3.61%), and others –including SINEs, DIRS, PLE, MITEs, and Maverick elements– (2.81%) to the total genome assembly.

https://doi.org/10.1371/journal.pone.0337390.g001

All superfamilies exhibited a wide variability in the size distribution of their TEs copies. For example, many copies of the DIRS and SINEs orders and the Loa superfamily were smaller than 500 bp, as well as some copies from the CMC, MULE-NOF, and Pogo superfamilies, which were less than approximately 200 bp (Figs S2 and S3 in S3 File). From a general perspective of the order level within LTRs and LINEs, most copies ranged from approximately 1 kb to 5 kb, with some Gypsy copies reaching up to 7.5 kb (Figs S2 and S3 in S3 File). Additionally, this size range was similar to that of the Helitron and TIRs copies, although for TIRs, a distribution starting from 500 bp was common (Figs S2 and S3 in S3 File). Finally, the Maverick superfamily had a few copies with large sizes, reaching up to > 8 kb, while the size of the MITEs was concentrated in a range of up to 3 kb (Figs S2 and S3 in S3 File).

Repeat landscape of transposable elements

The TE landscape illustrates the frequency at which full-length TE copies from each order were found at different divergence percentages compared to their consensus sequences from the manually curated TE library. A sustained increase in TE activity was observed at divergence percentages ranging from approximately 15% to 25%, with frequencies between 10,000 and 20,000 TE copies across the entire D. amaguana genome (Fig 2). This increase mainly involved TE copies from Helitron, TIR, and LTR orders, which may explain why these orders show a higher number of copies in the genome. Moreover, this pattern may reflect ancient transposition events, suggesting that many TEs have not been recently mobilized within the D. amaguana genome. Additionally, the peak of activity around 20% to 22% divergence could indicate a burst of transposition involving not only the aforementioned orders but also PLE and Maverick elements (Fig 2), which appear inactive in recent evolutionary time, as they show no distinguishable copies at lower divergence levels.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. TE landscape of the D. amaguana genome.

The landscape displays the distribution of full-length transposable elements (TEs) copies across divergence percentages, as a proxy for their relative age in the genome of D. amaguana. The x-axis represents the sequence divergence percentage of each TE copy from its consensus sequence, and the y-axis indicates the number of copies per TE order.

https://doi.org/10.1371/journal.pone.0337390.g002

De novo estimation of satDNAs in D. amaguana

Automatic annotation from RepeatExplorer analysis retrieved 65,528 clusters grouped into 65,514 superclusters. Among these, 23 clusters were identified as putative satDNAs, with four having high confidence and 17 having low confidence. After the examination, filtering, and final annotation of each putative satDNA, the number of confirmed satDNAs decreased to 16 (Table S5 in S2 File). Each cluster of these 16 satDNAs exhibited a globular or ring shape in the graph layout. In addition, they had C (connected component index) and P (pair completeness index) indices close to 1. These are key parameters for maintaining the identification of clusters as typical sequences of true satellites.

None of these satDNAs have been described to date. Moreover, BLASTN results did not show similarities with any satDNA families previously reported in 37 Drosophila species from the Sophophora and Drosophila subgenera [66] and/or with Drosophila species from the montium group [67]. Additionally, a homology search of the identified satDNAs did not yield any similarities with other satDNA sequences deposited in NCBI. Therefore, the consensus sequences for these 16 satDNAs were catalogued as new families, named DamaSat01–97 to DamaSat16–1260, according to the nomenclature rules proposed by Ruiz-Ruano et al. [68]. These complete sequences are available in S4 File.

DamaSat01–97 is a satellite with a total repeat length of 97 bp and the highest %AT content of 76.29%. Moreover, its contribution to the D. amaguana genome was 1.75%, which is a unique DNA satellite with a proportion higher than 1% in the genome. It is likely that DamaSat02–14 and DamaSat03–15 may be considered satellites with a medium proportion within the genome at 0.97% and 0.78%, respectively. The array lengths of these sequences were 14 bp and 15 bp, respectively, and they exhibited a similar %AT content of approximately 60%. Despite the consensus length of 14 out of 16 satellites up to 1.2 kb, the total array length for two sequences (DamaSat05–3461 and DamaSat11–6402) was significantly longer, approximately 3.4 kb and 6.4 kb, respectively. These sequences exhibited high satellite probabilities of 0.979, C and P indices close to 0.99, and a clear ring-graph layout. These characteristics are consistent with those typically associated with satDNAs, and the total array lengths support their classifications. Finally, we estimated the amount of satDNAs present in the raw reads of D. amaguana to be 4.90%.

Contribution of the repetitive DNA within the D. amaguana genome

Our analysis revealed that 61.76% of the assembled genome of D. amaguana consisted of nonrepetitive DNA (Fig 3A). The total content of repetitive DNA present in the large genome assembly of D. amaguana included transposable elements, satellite DNAs, and simple repeats (Fig 3A). Of these repetitive DNA, TEs were the most abundant type of repetitive sequences in this genome, accounting for 21.54%, which corresponded to approximately 98.1 Mb. Helitrons were the most abundant, with 29.50% of the total TEs, followed by LTRs (23.81%), TIRs (16.87%), and LINE (16.78%), while the rest of the TE orders contributed 13.03% (Fig 3B). Other types of repetitive DNA in the assembled genome of D. amaguana included simple repeats (11.80%). In contrast, satellite DNA may represent 4.90% of the genome based on raw reads.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Repetitive DNA.

(A) Estimated percentage of repetitive DNA comprising the large genome of D. amaguana calculated using de novo assembled genome. The satDNA percentage is an estimate based on raw reads (B) Percentage of each transposable element order relative to the total TE content annotated in the assembled D. amaguana genome after manual curation.

https://doi.org/10.1371/journal.pone.0337390.g003

Transposable elements within the D. amaguana genome

Genome size has been associated multiple times with TE content. This has led to the consideration that TEs are key actors in insect genomes [8], not only by their contribution to genomic structures, but also by their role in the evolution and adaptation of the species [11]. Within the insect order Diptera, TE content ranges from 1% to 55% [14]. In Drosophila species, TE content varies from 2.7% to 25% [71], even reaching up to 40% in species such as D. ananassae [14]. We assessed the percentage of TEs in D. amaguana and estimated that the mobilome corresponded to 21.54%. To our knowledge, this is the first report of TEs in a species from the mesophragmatica group, which has a large genome size (~455.5 Mb). Moreover, it could provide new insights into genome organization and potentially offer information on the role of these sequences in the adaptive radiation of Neotropical groups.

The best approach for de novo transposable element discovery and identification is still an active and unresolved research question [71]. The diversity of software and methodologies can generate slightly different results for the same species [72]. Although each pipeline has a defined workflow or shares tools to identify TEs, developers can define different approaches. Our preliminary results obtained through an automatic de novo TE annotation with three different tools showed that the TE content in D. amaguana ranged from 8.24% to 19.90%. This particularity has also been observed in D. willistoni, where two different bioinformatics tools estimated 9% and 16% of TEs in its genome [71], and in the genome of D. ananassae, which showed 12% and 20% of TEs when using the same tool but considering a reference library or with a de novo approach, respectively [71].

Although the development of tools for identifying and annotating mobile elements has been growing, characterizing the mobilome in a genome is still a challenge owing to the high complexity of repetitive DNA [73]. However, the raw TE libraries generated by de novo tools may contain redundant, fragmented, or false-positive TE consensus sequences [74]. For this reason, it must always be manually checked to ensure that the sequences are accurate and truly belong to TEs [55,56]. The manual curation of the D. amaguana mobilome included raw TE libraries from EDTA and RepeatModeler, excluding the results from reasonaTE because its estimated percentage was low and did not automatically provide a TE consensus library. Therefore, the analysis of D. amaguana identified 737 elements, 716 of which were discovered for the first time in Drosophila. This suggests that these sequences may represent novel TE families.

The TE annotation of the entire genome of D. amaguana performed with our curated library but also including curated libraries from other species of Drosophila, allowed us to gain insight into the different TEs types that compose this genome. One of the most relevant findings was the presence of the P superfamily, as the transposition of P elements has been associated for many years with hybrid dysgenesis events [27]. Within the mesophragmatica group, tropical species reported in Brazil, Colombia, and Chile, such as D. gaucha, D. gasici, D. pavani, and D. viracochi, did not show the presence of P elements [75,76]. Therefore, this finding would help to understand the role and origin of this superfamily in the host genome and the adaptive process, because the emergence of this class of TEs has been associated with horizontal transfer events [77]. It can occur owing to geographic, temporal, and ecological factors [78].

However, the detection of MULE-NOF and CMC superfamilies, although they comprise less than 0.01% of the genome, is still noteworthy. Despite some reports of Muta1 elements identified in insects such as Aedes aegypti [79], characterization of the MULE superfamily in other species is quite rare [80], as studies have focused more on fungi and plants [81]. A similar situation has been observed in the CMC superfamily. Although there are studies on CMC elements in insects, such as Mengenilla moldrzyki, Rhodnius prolixus, and Bombyx mori [82], these findings are limited because most reports focus on plants, which can account for up to 10% of transposons in the genome [83].

Many insects are susceptible to invasion by the Tc1-Mariner superfamily [84]. Some families of this class of TEs have been shown to be autonomously functional in Drosophila [85], even in evolutionarily distant species, thereby introducing genomic novelty [86,87]. Mariner elements have been identified in vitro for D. gasici, D. brncici, and D. viracochi [76]. Thus, D. amaguana was not far behind because the Tc1-Mariner superfamily was the highest contributor within the TIR order for the classified superfamilies, with an abundance of 0.48%.

LTR retrotransposons are a widely represented class of TEs in Drosophila genomes. D. yakuba, D. simulans, D. sechellia, and D. erecta have 141 new LTR elements and 76 sequences homologous to TEs of D. melanogaster have been described [88]. In this study, we annotated 295 LTR retrotransposons in D. amaguana and identified 192 new elements. Within this order, Gypsy and Copia were the superfamilies with the highest (25,029) and lowest (689) number of copies, representing 2.69% and 0.09% of the whole genome composition, respectively. A low number of copies in the Copia superfamily has also been observed in the model species D. melanogaster, where it has been relatively number compared to other organism classes, barely between 10 and 100 copies [89]. However, this superfamily is widespread in nature owing to its potential transposition activity, and some of them have been involved in horizontal transfer events between D. melanogaster and D. willistoni [88,90]. The Gypsy superfamily is highly diversified and numerous in several Drosophila species, including D. willistoni, D. ananassae, D. elegans, and –with the exception of D. erecta– all the species in the melanogaster group have been reported to contain between 41 and 61 different Gypsy families [88]. This makes it the most abundant superfamily of LTR retrotransposons in species, such as D. sechellia (9.36%), D. grimshawi (9.77%), and D. ananassae (12.44%), which contain Ty3/Gypsy type elements [88]. Additionally, D. brncici, D. pavani, D. viracochi, and D. gasici (with three populations from Chile, Colombia, and Bolivia) have been reported to contain Gypsy retrotransposons in their genomes [76]. Homologous sequences of Copia retrotransposons have been found in D. pavani, but there is limited information about Copia retrotransposons in other species of the mesophragmatica group [91].

Petersen et al. [14], assessed 62 arthropod genome sequences of insects identifying that LTR retrotransposons are dominant in the Diptera order, DNA transposons are representatives in Hymenoptera, LINE retrotransposons are prevalent in Hemiptera, and SINE retrotransposons had the lowest contribution in all insect order. The mobilome of D. incompta, a neotropical species with restricted ecology, belongs to the flavopilosa group that depends on flowers for its development [92], indicating that 14.62% of the genome corresponds to TEs, with LTR, LINE retrotransposons, and Helitrons as the most representative groups [73]. Other study in the genome of D. buzzatii st-1 (8.43% of TEs in a genome of ~153 Mb), D. buzzatii j-19 (4.15% of TEs in a genome of ~146 Mb), and D. mojavensis (15.35% of TEs in a genome of 194 Mb) identified Helitrons as the most abundant group in all species [93]. LTR retrotransposons were the second order more abundant only in D. mojavensis, due to the presence of LINE elements in both D. buzzatii species, whereas TIR transposons are catalogued as the third group abundant in all of them [93]. Our results in D. amaguana showed that Helitrons (6.35%), followed by LTR (5.13%), TIR elements (3.63%), and LINE (3.61%) were the most abundant TEs in genome composition. The abundance of Helitron elements has also been observed in D. virilis, D. miranda, and D. pseudoobscura [14]. SINE retrotransposons contributed to only 0.02%, which is commonly observed in most dipterans, where SINEs contribute less than 1% [14], or in the whitefly species Bemisia tabaci, where SINEs were the least represented TEs at 0.14% [94]. Furthermore, despite many Drosophila genomes showing less diversity and abundance of MITEs compared to other mosquito species and plants [95], D. amaguana has 1.17% MITE, making them the four most abundant class of TEs.

Analyzing the abundance of TE in D. melanogaster has also been attractive to some researchers with a predominance of LTR retrotransposons, followed by LINE elements, and finally, TIR transposons [58,96]. A conservative approach suggests that trend in the rank abundance of TEs in these orders is conserved across genus [71]. Usually, Class I elements predominant over Class II [8]. Nevertheless, there are exceptions that do not adjust to this postulate, such as D. erecta, which has a higher abundance of TIR elements, D. pseudoobscura, which has lower abundance of LTR retrotransposons [71], and now D. amaguana. All these variations and differences are owing to a variety of host factors, such as the transposition rate of progenitors, and the rate at their copies are inactivated and become unable to transpose [5].

TE landscape

Understanding the evolutionary dynamics of TEs is essential for comprehending when these sequences colonized the genome and how they contribute to expanding genome size. For this purpose, TE landscapes are very useful to visualize and characterize TE dynamics [73]. The graph groups TE copies based on the percentage divergences regarding their respective TE consensus sequence. Copies on the left of the graph correspond to recent copies, while those on the right correspond to more divergent copies compared to their consensus [97].

Almost all drosophilids within Diptera order, have shown a unimodal distribution with highest peaks corresponding to LTR retrotransposon as result of recent activity into the dipteran genomes or DNA transposons in ants [14]. In D. amaguana, the TE landscape shows a unimodal shape with an important TE activity ranging from 15% to 25% divergence. The TE activity refers to the active mobilization or continuous insertion of transposable elements in the genome, indicating that these elements are still active and dynamic. The main peak, defined between 20% and 22% sequence divergence, represents approximately 7–8% of the total number of Helitron copies. Furthermore, the occurrence of older transposition events involving Helitrons is well documented. For example, approximately 1% of the D. melanogaster genome consists of ancient Helitron remnants, while around 3% of the D. yakuba genome is composed of both autonomous and non-autonomous Helitrons, reflecting their historical activity within the genome.

In D. melanogaster, most LTR retrotransposons are considered relatively recent elements [98]. In contrast, in D. amaguana, several LTR copies (approximately 3,000–3,500) exhibited up to 20% sequence divergence from their consensus. This may suggest substantial TE activity, reflected by patterns indicative of older insertions, particularly for LTR retrotransposons and Helitrons. TIR elements also exhibit a strong signal at around 20–22% divergence, which may indicate a past burst of activity, although their apparent absence at lower divergence levels suggests they have not been recently active in D. amaguana. Such TE bursts may have resulted from successful habitat expansions, potentially linked to stress-induced disruption of protective mechanisms such as the piRNA pathway, or from interspecies contact [99]. For example, the well-known case of P element transfer from D. willistoni to D. melanogaster occurred after range overlap in South America; soon after, the P element rapidly spread across global D. melanogaster populations.

The low contiguity of the assembled genome may have hindered the detection of more recent insertions with low sequence divergence. However, given the limited genomic information available for D. amaguana, further analyses are needed to confirm or refine these observations.

Satellite DNA as other repetitive DNA contributors

As well as mobile elements, other types of noncoding DNA, may also be key to understanding why some species have large genome sizes. Transposable elements and satellite DNA are the main driving force influencing the genome size of Drosophila; these repetitive sequences have shown a positive correlation with larger genomes sizes [100]. For example, both longer introns [101] and lengthier stretches of microsatellites [102] have been reported in D. virilis compared to D. melanogaster, which could help explain why the increased presence of non-coding and repetitive DNA causes the D. virilis genome to be approximately twice as large as that of D. melanogaster genome. Another observed case is D. cyrtoloma (401 Mb), the species currently known to have the largest genome within the Drosophila genus. Its large genome is due to at least ~70% satellite DNA content [69]. However, there are also findings where satDNAs are not the principal contributors of repetitive DNA and genome size [103], such as in D. erecta (153 Mb), where only 0.4% of its genome is composed of satDNA [104].

The class Insecta includes a wide range of satDNA families [68,105–108]. These sequences are involved in evolution, being unique to each species and showing high dynamism within repetitive DNA [11,109]. For instance, up to 16 families of satDNAs have been reported in D. melanogaster [110], between 2–14 families in species from the montium group [67]. We found 16 satDNA families in D. amaguana, corresponding to 4.90% of the genome. This proportion is similar to the 5% satDNA content reported in D. simulans (142 Mb) [104]. None of the satDNA sequences from D. amaguana showed homology with families reported in other species, probably due to underrepresented sequences in available databases.

The satDNAs in Drosophila are variable in terms of repeat array size, nucleotide composition, and genomic proportion [66]. The variation in repeat length among the satellites found in D. amaguana ranges from 6 bp (DamaSat04–6 and DamaSat09–6) to 6.4 kb (DamaSat11–6402), which, to the best of our knowledge, is the largest satDNA reported in insects. This is significantly larger than previously reported satDNAs in other species, such as Chrysolina americana (3664 bp) [111] and Mahanarva species (4228 bp) [112]. Most common satDNA lengths in the Drosophila genus range from <10 bp to 400 bp [113–115], with some even reaching up to 570 pb [66]. 11 out of 16 satDNA families in D. amaguana fall within this range, while three families (DamaSat05–3461, DamaSat11–6402, and DamaSat16–1260) had a consensus length greater than 1 kb. Monomer sizes exceeding 1 kb have also been described in other insect species, such as the Vandiemenella morabine grasshoppers [116], the ladybird beetle Hippodamia variegata [117], the beetle Euchroma gigantea [118], and the red flour beetle Tribolium castaneum [108], where repeat lengths can extend to several thousand base pairs [119].

General overview of repetitive DNA content in D. amaguana

Initially, 21.54% of TEs reported in D. amaguana may seem low compared to other species with genome sizes up to three times smaller. For example, D. melanogaster has been reported to have 17.11% of TEs in an estimated genome size of 142.6 Mb [51]. This value was close to approximately 20% TE content reported for a 180 Mb genome [120], or the 19.3% reported by Mérel et al. [37]. Although this comparison is not intended as a definitive conclusion, it serves as a hypothesis based on preliminary data suggesting that the TE percentage may be relatively low. Nonetheless, it is important to acknowledge that, in many cases, there is a reported trend of a positive correlation between genome size and TE content, with larger genomes typically containing a higher proportion of TEs [8,121]. On the other hand, simple repeats content represented 11.80% of the genome, while satDNA content estimated using raw reads was 4.90%.

During the last two decades, multiple studies in Drosophila have contributed to new knowledge about the factors underlying genome organization. In this context, genome size in Drosophila has been approached from diverse perspectives. Ecological factors in the environments where species develop may be related to the genome size, as certain habitats could reduce the likelihood of horizontal transfer events involving TE families. For example, D. grimshawi was described as an island endemic species with low transposable element content [71]. Therefore, variation in TE content may result from adaptive responses to environmental stress during specific periods or within particular ecosystem [122]. This supports a hypothesis proposed years ago, suggesting that DNA content is often influenced by the environment in which an organism lives [123]. Other important contributors to genome size include chromosomal rearrangements [124], the presence of heterochromatic genes [125], and the role of other types of noncoding DNA [126,127]. Additionally, the percentage of duplicated genes observed in the assembled D. amaguana genome provides an opportunity to investigate the basis of this particularity in future studies. Duplication events in insects have been shown to play key roles in genome size variations [128], evolutionary innovative [129], phenotypic diversification [130,131].

However, it should also be noted that short-read assemblies often struggle to accurately resolve repetitive regions and highly similar sequences, leading to gaps, unassembled regions, fragmented contigs, collapsed repetitive sequences, and redundant or duplicated sequences in the final assembly. These issues can affect assembly accuracy and complicate downstream analyses, including gene prediction and the functional annotation of many genetic elements. Although the initial Redundans results suggested low heterozygosity and less than 2% duplicated contigs in our assembly, it is important to emphasize that the high level of gene duplication observed in the D. amaguana genome may or may not solely reflect biological reality. In addition to genuine gene duplications –which could be the focus of future specific studies in this species– technical limitations inherent to short-read sequencing may artificially inflate duplication estimates. Similarly, the estimated proportion of transposable elements (TEs) may also be underestimated, considering the generally positive correlation between TE content and genome size observed in other species or taxa. These results should therefore be interpreted with caution and will require confirmation and refinement using long-read sequencing technologies.

Further detailed research and deeper analyses may significantly help answer key questions about the larger genome size observed in D. amaguana. Expanding the study to include other individuals or populations [132], and applying third generation sequencing technologies to obtain long reads for generating a new genome assembly, could help address unresolved questions within the mesophragmatica group. These approaches may allow the detection of more complete sequences [133,134], overcoming current limitations not only to better characterize mobile elements [34,135], but also to provide a more accurate and comprehensive understanding of the biology of this Andean species of Drosophila. In particular, third-generation sequencing –which offers greater continuity and accuracy across complex genomic regions– will be essential to more reliably assess gene duplication, refine estimates of TE content, and ultimately clarify the true structure and composition of the D. amaguana genome.

In conclusion, our study showcases the first effort in the identification of repetitive DNA –including transposable elements (TEs) and satellite DNAs (satDNAs)– in Neotropical Drosophila species belonging to the mesophragmatica group that inhabit Andean forests in Ecuador. We identified 21.54% TEs and reported 11.80% simple repeats in the ~ 455.5 Mb draft genome assembly. In addition, we estimated the satDNA content using raw reads, which resulted in 4.90%. Our findings revealed the presence of all the TE orders, with a high prevalence of Helitron, TIR, LTR and LINE superfamilies. Moreover, we discovered 16 new satDNA families. The TE and satDNA content in D. amaguana do not appear to significantly contribute to its large genome size, as the TE proportion is similar to that of other Drosophila species with smaller genomes. However, these results provide valuable insights for future research on the role of mobile sequences and genome architecture in the adaptation and development of D. amaguana. They also represent a starting point for further analyses within the mesophragmatica group to better understand, among other things, the factors driving large genome sizes and evolutionary processes.