A reference mitogenome for the yellow-tailed woolly monkey (Lagothrix flavicauda) using Oxford Nanopore sequencing shows no evidence for adaptation to high elevations

Abstract

Background

Species inhabiting high-elevation environments face energetic and physiological challenges, creating opportunities for adaptive divergence from lowland taxa. The yellow-tailed woolly monkey (Lagothrix flavicauda), a Critically Endangered primate endemic to high-elevation forests in Andean Perú, provides a unique system for investigating the evolutionary consequences of long-term occupation of high elevations. We used Oxford Nanopore Technology (ONT) to generate a high-coverage mitochondrial reference genome using low-quality fecal samples. Using this and a new mitogenome from a closely-related lowland common woolly monkey (L. lagotricha poeppigii), we estimated divergence dates and calculated substitution ratios to investigate the evolutionary history of both Lagothrix taxa and assess whether high-elevation living has left signatures of selection in the mitochondrial oxidative phosphorylation genes of L. flavicauda.

Results

We assembled a high-coverage (9190X) reference mitogenome of 16,603 bp for L. flavicauda from a 15-year-old fecal sample, and a novel mitogenome for L. lagotricha poeppigii from fresh tissue (112X, 16,590 bp). Our phylogenetic reconstruction is consistent with other platyrrhine primate phylogenies with high bootstrap support, with the two Lagothrix taxa showing evidence of having diverged from one another ~ 2.8 million years ago. Substitution (d_N/d_S) ratios suggest purifying selection on protein-coding mitogenomic regions in all assessed lineages.

Conclusions

We found no evidence of positive selection acting on mitochondrial protein-coding genes in either L. flavicauda or the broader genus Lagothrix; all analyses instead support pervasive purifying selection across genes involved in oxidative phosphorylation. These results suggest that unlike in some other non-human primates occupying high elevations, the adaptive response of L. flavicauda to living in the Andes does not appear to involve evolution in the mitochondrial oxidative phosphorylation pathway, and likely involves alternative genetic pathways or mechanisms not captured by mitochondrial sequence evolution. By providing the first publicly available high-coverage mitochondrial genome for L. flavicauda, this study establishes a foundation for future population mitogenomic investigations of adaptation, evolutionary history, and conservation in this Critically Endangered, high-altitude primate.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12864-026-13250-4.

Background

High-elevation environments impose a unique suite of physiological and energetic challenges to organismal survival, including reduced oxygen availability, lower temperatures, increased solar radiation, and reduced ecosystem productivity [1–3]. Although atmospheric oxygen concentration remains relatively constant across elevations, declining atmospheric pressure reduces the partial pressure of oxygen and its availability to tissues, resulting in hypoxic conditions [4]. This reduced oxygen availability can limit ATP production through oxidative phosphorylation (OXPHOS), increase oxidative stress, alter metabolic processes, and negatively affect reproductive performance and other fitness-related traits in individuals not adapted to such environments [2, 5–8]. Consequently, populations that persist at high elevations frequently exhibit physiological, behavioral, and genetic modifications to improve oxygen utilization and energy production. Understanding the evolutionary mechanisms underlying these responses has become a major focus of studies investigating adaptations to extreme environments.

The Andes of South America provide a particularly valuable system for studying adaptation to high elevations, as they encompass steep environmental gradients over relatively short geographic distances and support exceptional levels of biodiversity and endemism. Elevational differences in climate, vegetation, productivity, and oxygen availability create strong ecological boundaries that restrict dispersal and promote divergence among populations [9–11]. These effects may be especially pronounced in tropical mountains, where species often exhibit narrower environmental tolerances and lower dispersal across elevational zones than their temperate counterparts [9, 10]. As climate change drives upslope shifts in suitable habitat and alters competitive interactions among species, understanding the genetic basis of adaptation to Andean environments is becoming increasingly important for predicting species persistence and responses to environmental change [12–14].

The yellow-tailed woolly monkey (Lagothrix flavicauda)—a Critically Endangered platyrrhine primate endemic to the yungas cloud forests on the eastern slopes of Andean Perú [15, 16]—provides a compelling system in which to investigate the evolutionary consequences of long-term occupation at high elevations in the Andes. It is known to inhabit an altitudinal range of 1500 to 2800 masl [17, 18], although in some regions it has been observed as low as 1000 masl [19, 20]. The species was re-documented in 1974 after having been thought extinct for half a century [21]. Despite difficulties finding and following the monkeys in their montane habitat, it has experienced a steady increase in ecological and behavioural research in the past two decades, albeit with the inclusion of few genetic studies. Early phenetic and morphological studies resulted in an unclear taxonomic placement for the species, with some arguing it belonged to the monotypic ateline genus Oreonax [22, 23]; mitogenomic variation in L. flavicauda, however, is similar enough to that of lowland woolly monkeys to place all woolly monkey taxa in the genus Lagothrix, with all lowland taxa placed within L. lagotricha as a sister species to L. flavicauda [24]. Behavioral data from the sole long-term study of the species suggest that L. flavicauda is frugivorous but tends to eat a lower diversity of plants and a higher proportion of insects than more well-studied lowland woolly monkey species [25–27], despite having similar group sizes and population densities [28–31]; it is yet unclear if these dietary differences are in response to the physiological/energetic stresses of living at high elevations (as has been proposed for a similar shift in diet among a high elevation population of L. lagotricha lugens [32]) or a regional artifact unique to the single population that has been extensively studied. Moreover, the rapid degradation of yungas habitat and hunting [33] by humans in Perú, have made L. flavicauda a particular conservation concern. Enhancing the available genetic information for the species can benefit conservation efforts by providing insight into population structure [34–36], the extent of inbreeding [36], and remaining genetic diversity [37].

While there are other woolly monkey taxa that occupy both low and high elevations, L. flavicauda is the only woolly monkey to be restricted to the high montane forests of the Andes. The evolutionary mechanisms underlying the differences in L. flavicauda ecology (diet, morphology, habitat use) relative to other woolly monkeys are poorly understood and have yet to be interpreted through the lens of altitude differences. The evolutionary history of the genus further complicates this picture, as some authors have suggested that Lagothrix may have emerged as a montane genus during the Andean uplift prior to the expansion of common woolly monkeys (L. lagotricha) into the Amazonian lowlands [38] which raises the possibility that adaptations to montane environments predate the divergence of common and yellow-tailed woolly monkeys. Genetic data provide a valuable means of evaluating these alternative scenarios by allowing the direct investigation of evolutionary patterns associated with the occupation of high-elevation habitats.

Blood and tissue samples are the gold standard for providing high-quality DNA at sufficient quantities for genome-level sequencing (see [39] for sample material comparisons) but necessitate invasive sampling. Although a small number of L. flavicauda individuals have been rescued from the pet trade and maintained for a short time in captivity, they were not a part of managed breeding or research programs and biological tissues suitable for genomic analyses were not generally collected or archived. Museum specimens are relatively rare for this species, and obtaining genomic-quality DNA can be challenging due to degradation and restrictions on destructive sampling [40]. Further, even if their Critically Endangered status did not make the invasive sampling of wild individuals non-viable, the eastern Andean terrain has made L. flavicauda exceptionally difficult to locate, follow, and especially collect biological samples from in the wild (but see [41] for details on one successful long term research site). Consequently, tissue-derived genetic resources are extremely limited, and fecal samples remain the most feasible – and indeed only – means of acquiring genetic data for this species. Fecal samples can be an excellent source of genetic information, particularly for organisms associated with the host (i.e., various microbiomes or parasites; [42–44]), and third-generation sequencing is increasingly used for such data acquisition [45]; high quality host DNA, by contrast, can be difficult to obtain from fecal samples due to DNA degradation and the relatively low proportion of host to microbial and digested plant DNA. Targeting the host mitochondrial genome (mitogenome) increases the chances of sequencing host DNA due to its natural abundance (1–10 thousand copies of the mitochondria per cell) compared to the nuclear genome [46].

The mitogenome is completely separate from and has a much higher mutation rate than the nuclear genome [46]. Due to its relatively high mutation rate and non-recombining mode of inheritance the mitogenome is a widely used system for numerous evolutionary genetic questions, including phylogenetic reconstruction, assessment of lineage divergence, taxonomy/systematics, and relatedness estimation, among other general population genetic analyses [47–49]. However, because the mitogenome is maternally inherited and effectively represent a single locus, evolutionary patterns inferred from mitochondrial data do not necessarily reflect those present across the nuclear genome, and phylogenetic relationships inferred from mitochondrial sequences may differ from those inferred from other independent loci (e.g. [47]). Despite these limitations, mitochondrial data have proven particularly valuable for taxa that are difficult to sample and for which genomic resources remain limited.

To our knowledge, previous mitogenetic studies of L. flavicauda have exclusively focused on comparative work to establish L. flavicauda taxonomy or their placement in larger platyrrhine primate systematics. For example, Di Fiore et al. [24] generated several (unpublished) mitochondrial genomes using Sanger sequencing to evaluate the position of L. flavicauda within the atelines, concluding that they fall within the genus Lagothrix (rather than their own separate genus, Oreonax), while Ruiz Garcia et al. [38] included L. flavicauda in comparative work using fragments of MT-CO1 and MT-CO2 to explore the taxonomic placement and diversification of L. lagotricha tschudii within Lagothrix. Despite its recent inclusion in Lagothrix-focused comparative work, some population mitogenomic analyses [50] and many larger-scale studies – such as recent work sequencing whole mitogenomes from over 200 platyrrhine taxa to better understand diversification within the Infraorder [51] – exclude L. flavicauda, likely due to the rarity of high-quality samples and publicly-available sequence data. Consequently, relatively little is known about patterns of mitochondrial evolution in L. flavicauda beyond its phylogenetic placement, including whether association with montane habitats has influenced patterns of selection in mitochondrial genes. Developing a reference mitogenome as a basis to search for selection therefore represents an important step toward addressing both conservation and evolutionary questions in this species [52].

Beyond their utility for phylogenetic and genetic analyses, mitochondrial genomes are particularly relevant to questions of high-elevation adaptation due to the mitochondria’s role in energy production via OXPHOS [53]. Given that all protein-coding regions of the mitogenome are involved in OXPHOS pathways, comparative analyses of substitution rates in the mitogenome across closely related taxa have been widely used as a first step in investigating whether adaptation to montane environments has left detectable signatures in these genes. Such methods, including d_N/d_S (calculating the ratio of nonsynonymous to synonymous mutation ratio), have inferred evidence of mitogenomic adaptation to high elevations in a variety of taxa (e.g., amphibians [54]; grasshoppers: [55]; Andean waterfowl: [56]). While such studies in non-human primates have been more limited, work in Chinese snub-nosed monkeys (Rhinopithecus spp.) found sites in the mitochondrial genes MT-ND2 and MT-ND6 of a high-elevation species (R. roxellana) to be under positive selection relative to closely related congeners residing at lower elevations [57]. This and many other studies involve species at elevations higher than that of L. flavicauda (> 3000 masl); however, ~ 2000 masl is considered moderately high or high elevation, with reports of lower hemoglobin oxygen saturation in individuals at this altitude compared to low altitudes or sea level due to the reduced partial pressure of oxygen, making it less available to tissues [58, 59]. The differences in oxygen availability are enough to induce physiological changes and stressors [4], which certain point mutations in the mitogenome have been shown to mitigate [60]. However, because the mitochondrial genome is a single maternally inherited locus, it captures only a portion of the genetic basis of adaptation. Thus, the absence of positive selection in mitochondrial genes does not necessarily indicate an absence of adaptation to high elevations. Nevertheless, no study has evaluated whether mitochondrial genes in Lagothrix exhibit signatures of selection associated with long-term occupation of montane habitats, making mitogenomic analyses a useful first step toward addressing this question.

The restricted distribution of L. flavicauda to mid- and high-altitude forests raises the question of whether prolonged occupation of higher elevations has left detectable signatures of selection in areas of the genome relevant to phenotypes adaptive to this environment, such as energy production through oxidative phosphorylation. Addressing this question first requires the development of genomic resources; although mitogenomic sequence data for L. flavicauda have been generated previously [24], these data are not publicly available as a high-coverage reference for the species. Further, this previous consensus sequence was generated using Sanger sequencing from fragmented PCR-amplified fecal samples from multiple wild individuals, which can present the problem of interpreting separate unassociated variants from different individuals in selection analyses. The first goal of this study was to generate a high coverage reference mitogenome for L. flavicauda using DNA from noninvasively collected fecal samples using third-generation Oxford Nanopore Technologies (ONT) sequencing. Our use of ONT sequencing combines portability and accessibility with the ability to generate long-read sequence data, making it a valuable tool for assembling complete mitochondrial genomes and expanding genomic resources for non-model and conservation-relevant species, building upon previous ONT-based non-human primate mitogenome sequencing methods and illustrating its utility for working with noninvasively collected low-quality samples [61–66]. In addition to the new reference mitogenome for L. flavicauda, we generated a reference mitogenome for a closely related lowland woolly monkey subspecies, Poeppig’s woolly monkey (L. lagotricha poeppigii) for use as another Lagothrix species in phylogenetic and selection analyses. We then reconstructed phylogenetic relationships among atelid primates, both to establish the evolutionary context of the newly generated mitogenomes and to provide the phylogenetic framework required for subsequent codon-based analysis of selection. Using these data, we estimated d_N/d_S ratios for mitochondrial protein coding genes—all of which are important to the OXPHOS pathway—to test two alternative hypotheses regarding the evolutionary history of high-elevation adaptation in L. flavicauda. If adaptation to montane environments evolved specifically within the L. flavicauda lineage following its restriction to high forests, we expected the preferred d_N/d_S model to reveal signatures of positive selection on the L. flavicauda branch. Alternatively, if adaptation to montane environments arose earlier in the evolutionary history of Lagothrix, signatures of purifying selection will be detectable across the broader Lagothrix lineage relative to other atelids, as shown in other species originating in high altitudes [56]. Finally, we used these and other primate mitogenomes to estimate divergence times within Atelinae to provide temporal context for the diversification of Lagothrix and the evolutionary history of habitat occupation within the clade.

Methods

Sample collection and extraction

We non-invasively collected L. flavicauda fecal samples from wild individuals at a research area locally known as Peroles, the site of the only published long-term study of the species [27, 41], located in the Comunidad Campesina de Yambrasbamaba in the Región Amazonas, Perú. The site is located on the eastern slopes of northern Andean Perú and contains disturbed primary forest and regenerating secondary forest interspersed with pasture [28, 67]. We collected the samples during an expedition to the region in May 2008 with the support of the NGO Neotropical Primate Conservation and Universidad Nacional Mayor de San Marcos in Lima, and we stored the samples in commercial RNAlater buffer at a ~ 1:1 ratio of preservative to feces. We kept samples at room temperature for up to six months and then stored them long-term in a -80° C freezer. The age and lack of consistent cold chain storage conditions of these samples establish the low-quality potential for host DNA yield.

We extracted DNA from the samples in 2023 (15 years after collection) using the QiaAMP Fast DNA Stool Mini Kit following the manufacturer’s protocol, with all subsequent modifications adopted from Di Fiore et al. [24], which optimized fecal DNA extraction for frugivorous primates: Despite more recent kits excluding Buffer ASL, we purchased the buffer separately and used it as recommended in the modified 2010 kit protocol. After the addition of Buffer ASL, we incubated samples at 56° C for 30–60 min and for an additional 30–60 min at room temperature on a rocking platform. We added 500 uL of InhibitEX, rather than an InhibitEX tablet as recommended in the modified 2010 kit protocol, and vortexed immediately. We also heated the elution buffer to 70° C by incubating for 5–10 min. We pipetted 100 uL of the elution buffer directly onto the spin column membrane and allowed it to incubate at room temperature for 30–45 min before final centrifugation to maximize elution. We measured concentrations of extracted DNA using a Quantus Flourometer and QuantiFlour dsDNA system.

To provide a closely-related lowland comparison for phylogenetic and selection analyses, we also generated a mitogenome for L. lagotricha poeppigii from ONT sequencing. We utilized a tissue sample that was opportunistically collected from a wild-born L. lagotricha poeppigii infant who died after falling from its mother at the Tiputini Biodiversity Station (TBS) in Ecuador. A 200 mg tissue sample from this individual was immediately processed and sequenced in situ at the field station after extracting DNA using a Qiagen DNeasy Blood and Tissue kit. For all extractions, a negative extraction control consisting of molecular-grade water was included in the extraction process to monitor for potential contamination during processing.

All collections and research adhered to the legal requirements of the countries from which the samples were collected (Ecuador and Perú) and were carried out in compliance with The Convention on International Trade in Endangered Species of Wild Flora and Fauna (CITES). Sample collection for L. flavicauda was approved by the New York University IACUC, and sample collection for L. lagotricha poeppigii was approved by The University of Texas at Austin IACUC. All animal handling and interactions are consistent with the ethical recommendations of the American Society of Primatologists.

Mitogenome enrichment and sequencing

Since the majority of fecal DNA is non-host, we enriched the lower-quality L. flavicauda samples to isolate mitochondrial DNA using PCR with two overlapping long-range primer pairs that cover the entire mitogenome. These primer pairs, platy3 and platy4, cover approximately 9,800 and 11,650 bp, respectively, and both have been used previously to amplify platyrrhine mitochondrial DNA [24]. For separate amplification using each primer pair, we used 2 uL of DNA template in a 20 uL PCR reaction including 1X SuperFi II buffer containing 1.75 mM MgCl₂, 1X SuperFi II DNA polymerase, 200 μm dNTPs, 0.5 μm each of forward and reverse primer, and the rest of the volume in water. Cycling conditions remained the same as previously published [24]: for platy3, samples underwent an initial denaturation at 98° C for 2 min, followed by 40 cycles of denaturing at 98° C for 10 s, primer annealing at 60° C for 30 s, and extension at 68° C for 9–11 min, with a final extension at 68° C for an additional 9–11 min. For platy4, samples underwent an initial denaturation at 93° C for 3 min followed by 10 cycles of denaturing at 93° C for 15 s, annealing at 60° C for 30 s, and extension at 68° C for 11.5 min. This was followed by an additional 27–30 cycles of denaturation at 93° C for 15 s, annealing at 55° C for 30 s, and extension at 68° C for 11.5 min initially plus an additional 20 s for each cycle and a final extension at 68° C for 11.5 min. Negative PCR controls containing molecular-grade water in place of DNA template were included to detect potential contamination during PCR. We determined successful amplification through visualization after electrophoresis on a 0.5% agarose gel.

We prepared all positive amplicons (18 total from seven individuals) in a library using the ONT Native Barcoding Kit 24 V14 (SQK-NBD114-24) following the manufacturer’s protocol. We sequenced the samples on an ONT MinION Mk1B sequencer on a FLO-MIN114 flow cell with pore chemistry R10.4.1 for approximately 50 h. To further enhance the number of mitochondrial DNA reads, we used adaptive sequencing—a real-time selective sequencing approach in which molecules are compared to a reference sequence to reject non-target regions, therefore increasing the proportion of target reads. We used a high-quality white-bellied spider monkey mitogenome reference (Ateles belzebuth; NC_019800.1) as our target template because it represents a closely related ateline while remaining external to the focal taxa, which is important in allowing initial assembly and comparison of the Lagothrix mitogenomes without relying on a congeneric reference. While sequencing, live basecalling was performed using the fast-basecalling method embedded in Guppy v6.5.7, with reads later re-basecalled using simplex super accuracy (SUP; dna_r10.4.1_e8.2_400bps_sup@v4.3.0) with Dorado (v1.0.0). To generate the reference mitogenome, we chose one individual with both a high number of reads for each amplicon and minimal coverage imbalance between fragments. This avoids the creation of a chimeric reference from multiple individuals while minimizing low coverage errors. Genomic DNA from the L. lagotricha poeppigii tissue sample was prepared in a single library using the ONT Ligation Sequencing Kit (SQK-LSK114) and sequenced on a MinION Flow Cell with R9.6 pore chemistry for ~ 22 h at TBS.

Mapping, assembly, and visualization

We filtered all sequencing reads (from the platy3 and platy4 amplicons from a single L. flavicauda fecal sample and reads from the L. lagotricha poeppigii tissue sample) for quality (minimum Q15) and trimmed remaining adapters and PCR primers using fastp v0.23.4 [68]. With these cleaned sequencing data, we then constructed our reference mitogenomes for both species using a mapping and assembly pipeline adapted from Wanner et al. [66]: we first aligned reads to a high-quality A. belzebuth mitogenome (NC_019800.1; the same used as template for adaptive sequencing of the L. flavicauda sample) using minimap2 v2.24 [69], resulting in mapped reads that we then sorted, indexed, and converted to fasta format using samtools v1.21 [70]. We then assembled the mitogenomes in a two-step process using the resulting fasta file of aligned reads. We first used Flye v2.9.2 [71] for draft assembly and the first round of polishing, but due to its expected error rate of 0.5-1% for ONT reads, we followed this with a second polishing step using Medaka v1.0.3. We aligned the resulting consensus fasta files for each amplicon to the A. belzebuth reference in Geneious Prime 2025.1.2 (https://www.geneious.com) to create a single consensus whole mitogenome, which we annotated using MITOS2 v2.1.7 [72], and then converted to GenBank format for visualization and editing in Open Vector Editor v1.5.5.

Phylogenetic inference

To confirm the phylogenetic placement of our constructed mitogenomes and provide a tree for use in the selection analysis, we concatenated and aligned them with published whole mitogenome sequences from five other ateline taxa and one non-ateline platyrrhine outgroup (Aotus azarai) downloaded from NCBI GenBank (Table 1), using MAFFT v7.490 [73]. We chose only a limited number of species, as the selection models that utilize the resulting phylogenies perform more accurately using taxa closely related to the species of interest [74]. As later selection analyses require an alignment of only protein-coding genes, we inferred phylogenies using alignments of both complete mitogenome sequences of only the protein-coding genes (PCGs) to confirm consistent tree topology. Using these alignments, we reconstructed phylogenies using two methods—maximum likelihood (ML) and Bayesian—with three different partitioning schemes: a uniform substitution model across all genes, partitioned by gene, and partitioned by gene and codon (each codon treated as distinct partitions). For the uniform analyses, we chose the GTR + G4 + I substitution model for the whole mitogenome and all PCGs together based on model fit assessed via AICc in modelTest [75, 76]. For the partitioned analyses, we chose separate models for each partition via AICc using ModelFinder implemented in IQ-tree2 v2.3.6 [77]. The same model per gene was assumed when partitioning by gene and codon.

Table 1.

Mitogenomes used in the phylogenetic inference and selection analysis

Species	ncbi reference number	Genbank Accession Number	Reference
Brachyteles arachnoides	-	JX262672	[78]
Alouatta caraya	NC_064185.1	OM328861	[51]
Alouatta palliata	NC_064193.1	OM328926	[51]
Aotus azarai	NC_021939	KC757385.1	[79]
Ateles belzebuth	NC_019800.1	FJ785422	[80]
Ateles chamek	NC_064190.1	OM329060	[51]
Aotus azarai	NC_021939	KC757385.1	[79]
Lagothrix flavicauda	-	PZ228854	this study
Lagothrix lagotricha poeppigii	-	-	this study

We conducted ML analyses using RAxML v8.2.12 [81] with 25 random and 25 parsimony-based starting trees and 1,000 bootstrap replicates. Reliability of each node on the resulting ML tree are indicated by bootstrap percentages. For the Bayesian analyses, we used MrBayes v3.2.7a [82] to implement a Markov chain Monte Carlo (MCMC) method to infer phylogenies and achieve convergence of two independent, simultaneous runs, each consisting of five million generations and sampling occurring every 1,000 generations. A consensus tree was obtained after the initial 25% of trees were discarded as burn-in in each MCMC run. We compared phylogenetic inferences under the different schemes and inference methods to assess robustness of the resulting topology, which would be used for subsequent selection analysis. In the case that topology was different, we chose the preferred partitioning scheme and resulting tree using the AICc (for ML inferences) and Bayes Factor (BF, Bayesian inferences) and visualized our final trees using FigTree v1.4.4 [83].

Selection analysis

To search for the presence of positive selection in the mitogenome we calculated the non-synonymous to synonymous substitution ratio ω (d_N/d_S) within the mitochondrial PCGs of the selected platyrrhine lineages using the likelihood approach implemented in codeML, in PAML v4.10.7 [84]. With the PCG alignment in phylip format and the selected tree in newick format as input files, we conducted three sets of models to estimate ω: branch models, branch-site models, and site models. For the branch models, we used a one-ratios model (M0) as a null model as it assumes the same ratio for all lineages and therefore no variation in selective pressure [74, 85]. The free-ratios model (M1) assumes a different ω for each branch [85]. With an a priori prediction of positive selection in the L. flavicauda lineage due to high elevation adaptation, we used a two-ratios model (M2) to assign L. flavicauda as the foreground branch and the remaining lineages as the background branch, calculating two ω values for the phylogeny [74, 85]. We used likelihood ratio tests (LRT) to compare the null model (M0) to the two alternative models (M1 and M2) [74, 86]. To explore the possibility of the genus Lagothrix adapting separately from other platyrrhines, we repeated this analysis, running the M2 model with the whole Lagothrix clade as the foreground branch. Site models assume variable ω values among codons [87, 88], allowing us to infer selection in specific mitogenomic regions. Two nested pairs of models are run simultaneous in a single run in codeml: M1a (Nearly Neutral) vs. M2a (Positive Selection) [87, 88] and M7 (beta) vs. M8 (beta& ω) [84]. We again used LRT for each pair of site models to determine the best model [89].

In cases where positive selection was inferred for any lineage, we used the branch-site Model A to identify sites where ω > 1, indicating positive selection. This model considers two different ω values for the foreground and background branches along with values over amino-acid sites [90] to pinpoint potential areas of selection. Under this model, the codeML function uses the Bayes Empirical Bayes (BEB) approach [88] to calculate the posterior probabilities of specific codon sites, producing proportions of sites that are under positive or purifying selection in the foreground and background branches as well as identifying those most likely to be under positive selection. This was compared to a null version of this model with a fixed foreground ω = 1 using a chi-squared tested of the LRT statistic [74].

Divergence date estimation

Using our mitogenomic sequences, we sought to estimate the timing of divergence between Lagothrix and other ateline taxa using the Bayesian MCMC method in the program BEAST2 v2.7.7 [91]. We incorporated the whole mitogenome sequences of 42 primate taxa downloaded from GenBank in addition to our newly constructed references (Table 2). We extracted, aligned, and concatenated the PCGs of each mitogenome to create a 11,678 bp supermatrix partitioned by codon. We employed a relaxed lognormal molecular clock to allow substitution rates to vary independently across branches, and a GTR + G+I substitution model with 4 rate categories to account for among-site rate variation. We used a Yule pure birth process, which is appropriate for an interspecies dataset because extinction rates are difficult to estimate reliably from molecular phylogenies. We assigned the mean of the clock rate a lognormal prior and the birth rate parameter a Gamma shape prior.

Table 2.

Mitogenomes used in the divergence dating analysis

Parvorder	Family	Species	Accession number	Genbank Accession Number	Reference
Platyrrhini	Cebidae	Aotus azarai	NC_021939	KC757385.1	[79]
		Aotus lemurinus	NC_019799	FJ785421	[80]
		Cebus albifrons	NC_002763	AJ309866	[92]
		Oedipomidas oedipus	-	FJ785424	[80]
		Saguinus oedipus	NC_021960	KC757409	[79]
		Saimiri boliviensis	NC_021966	KC959987	[79]
		Saimiri boliviensis boliviensis	NC_018096	HQ644339	[93]
		Saimiri oerstedii citrinellus	-	HQ644336	[93]
		Saimiri sciureus	-	FJ785425	[80]
		Sapajus apella	NC_016666	JN380205	[94]
		Sapajus xanthosternos	NC_021961	KC757410	[79]
	Callitrichidae	Callithrix geoffroyi	NC_021941	KC757388	[79]
		Callithrix jacchus	-	AB572419	[95]
		Cebuella pygmaea	NC_021942	KC757389	[79]
		Leontopithecus rosalia	NC_021952	KC757399	[79]
	Pitheciidae	Chiropotes albinasus	NC_021946	KC757393	[79]
		Cacajao calvus	NC_021967	KC959985	[79]
		Plecturocebus cupreus	NC_021965	KC959986	[79]
		Plecturocebus donacophilus	NC_019801	FJ785423	[80]
	Atelidae	Alouatta caraya	NC_064185.1	OM328861	[51]
		Alouatta palliata	NC_064193.1	OM328926	[51]
		Aloutta seniculus	NC_064188	OM329058	[51]
		Lagothrix flavicauda	-	-	this study
		Lagothrix lagotricha poeppigii	-	-	this study
		Lagothrix lagotricha	NC_021951	KC757398	[79]
		Ateles belzebuth	NC_019800.1	FJ785422	[80]
		Ateles belzebuth	-	KC757386	[79]
		Ateles chamek	NC_064190.1	OM329060	[51]
		Brachyteles arachnoides	-	JX262672	[78]
Catarrhini	Cercopithecoidea	Chlorocebus aethiops	NC_007009	AY863426	[96]
		Colobus guereza	NC_006901	AY863427	[96]
		Theropithecus gelada	NC_019802	FJ85426	[80]
		Trachypithecus obscurus	NC_006900	AY863425	[96]
		Papio hamadryas	NC_001992	Y18001	[97]
		Macaca mulatta	NC_005943	AY612638	[98]
	Hominoidea	Gorilla gorilla	NC_001645	D38114	[99]
		Homo sapiens	-	EF061150	[100]
		Hylobates lar	NC_002082	X99256	[101]
		Pan troglodytes	-	EU095335	[102]
		Pongo abelii	NC_002083	X97707	[103]
Tarsiformes		Carlito syrichta	NC_012774	AB371090	[104]
		Cephalopachus bancanus	NC_002811	AF348159	[105]
Lemuriformes		Eulemur rufus	NC_021948	KC757395	[79]
		Lemur catta	NC_004025	AJ421451	[106]
Lorisiformes		Galago senegalensis	NC_012761	AB371092	[104]
		Nycticebus coucang	NC_002765	AJ309867	[92]

Calibration priors were selected to provide temporal constraints across both deep primate divergences and the platyrrhine/atelid ingroup, while prioritizing calibrations most relevant to the focal clade. All calibrations were ultimately derived from fossil-based age constraints previously used in primate divergence dating studies (Table S1). Most calibration priors were adapted from Di Fiore et al. [24], with minimum age offsets updated according to the fossil calibration review of de Vries and Beck [107] where applicable. The original calibrations of Di Fiore et al. [24] were themselves based on fossil constraints compiled by Chiou et al. [93]. For these nodes, prior distribution types and associated parameters (e.g., means and standard deviations) were retained from previous studies to maintain comparability with earlier analyses, whereas additional calibration points from de Vries and Beck [107] were implemented as published. Because our taxon sampling was designed to resolve the phylogenetic placement, divergence history, and molecular evolution of Lagothrix rather than to comprehensively estimate divergence times across Primates, divergence date estimates are interpreted primarily as a temporal framework for relative divergences within Platyrrhini and Atelidae. Consequently, absolute age estimates for deeper primate nodes should be interpreted with caution.

We approximated posterior distributions of parameters by sampling from 4 independent MCMC runs, each running for 25 million generations with parameters and trees sampled every 1,000 generations. We converged trees using LogCombiner v2.7.7 after discarding the first 20% of trees as burn-in, resulting in 80,000 remaining trees that were analyzed together in Tracer v1.7.2 [108]. We ultimately generated a consensus phylogenetic tree using TreeAnnotator v2.7.7 with node height distributions visualized in FigTree v1.4.4 [83].

Results

ONT sequencing

Extractions of fecal DNA for all L. flavicauda samples yielded between 3.8 and 39 ng/uL. We attempted to concentrate high-yield samples (> 10 ng/uL) that could fulfill the 1,000 ng recommendation by ONT for sequencing gDNA, but sequencing in this way provided no reads that mapped to the mitochondrial reference, as over 95% of sequences were bacterial DNA or unidentifiable. Given this, we changed our approach to successfully amplify fecal DNA from multiple individuals for each overlapping mtDNA primer pair (18 amplicons from seven individuals, with some individuals being amplified twice) with amplicon concentrations ranging from 2.11 to 46 ng/uL. We included all 18 amplicons in library preparation, resulting in a final library concentration of 6.1 ng/uL and approximately 10.5 fmol sequenced for approximately 50 h. Even though the amplicons were normalized after barcoding and before being pooled, sequencing depth varied among individuals and amplicons. We chose a sampled individual with the high available mitogenomic reads and minimal fragment coverage imbalance to construct the reference mitogenome (sample 08-138, adult female). This sample had an initial concentration of 15 ng/uL and a concentration of 46 ng/uL and 15 ng/uL after being amplified for the platy3 and platy4 fragments, respectively (Table S2). After being re-basecalled with SUP and trimmed, the barcoded platy3 and platy4 amplicons yielded 7,678 and 20,634 reads, respectively, a high percentage of which mapped to the A. belzebuth reference (Table 3). After assembling these fragments, they produced amplicon consensus sequences with very high mean coverage over all sites (4960X for platy3 and 12799X for platy4; Table 3 and Fig. 1). After combining and annotating the fragments in Geneious Prime 2025.1.2 (https://www.geneious.com), we produced a 16,603 bp mitochondrial genome with no gaps and a weighted mean coverage of 9190X (Fig. 2).

Table 3.

Total L. lagotricha poeppigii and L. flavicauda sequencing reads and coverage when mapped to the A. belzebuth mitogenome reference

	L. flavicauda - platy3			L. flavicauda - platy4			L. lagotricha poeppigii
	Total Raw Reads	Total Reads After Trimming	Aligned Reads	Total Raw Reads	Total Reads After Trimming	Aligned reads	Total Raw Reads	Total Reads After Trimming	Aligned Reads
Read count	16,866	7,678	6,474	34,836	20,634	15,158	666,112	556,251	2,049
Total (Mb)	134	51	50	309.6	156.4	151.4	2,380	1,981	3.3
Average length (bp)	7,936.70	6,597.50	7,672.20	8,988.1.60	7,577.60	9,987.30	3,573.70	3,560.70	1,614.30
% mtDNA	-	-	84.32%	-	-	73.46%	-	-	0.004%
Alignment coverage (mean)	-	-	4960X	-	-	12799X	-	-	112X

Fig. 1.

Depth of coverage for the (A) combined L. flavicauda sequencing reads and (B) L. lagotricha poeppigii reads across the A. belzebuth reference

Fig. 2.

Annotated assembly of the L. flavicauda and L. lagotricha peoppigii mitochondrial genomes

The sequencing run of gDNA extracted from the L. lagotricha poeppigii tissue sample produced 666,112 reads after being re-basecalled with super accuracy, 556,251 of which were left after trimming and filtering for quality. Of these reads, 0.368% mapped to the A. belzebuth reference mitogenome, creating a 16,590 bp mitogenome with 112X coverage (Figs. 1 and 2).

Phylogenetic reconstruction and selection analysis

We reconstructed phylogenies using three partitioning schemes with ML and Bayesian methods for alignments of whole mitogenomes as well as for just the PCGs. Models for each gene are given in Table S3 (whole mitogenome) and Table S4 (PCGs). Partitioning substitution models by gene and codon were strongly preferred in the ML phylogenetic inference, according to the log likelihood and AICc (Table S5). The Bayesian inference also showed preference to the model partitioned by gene and codon due to the lowest log likelihood leading to the highest Bayes factor (Table S5). The topologies of all resulting best trees were the same and all nodes had high support, but branch lengths differed between reconstructions (Fig. 3). As expected, all phylogenies place L. flavicauda in a clade with L. lagotricha poeppigii, and sister to the woolly spider monkey, Brachyteles arachnoides.

Fig. 3.

Phylogenetic inferences of ateline mitogenomes. Trees are resulting from analyses partitioned by gene and codon using Bayesian assessments of the (A) whole mitogenome and (B) just PCGs, and maximum likelihood (ML) assessments of (C) the whole mitogenome and (D) PCGs. Node labels for the ML trees represent percent bootstrap support while those on the Bayesian trees represent posterior probabilities. Asterisk (*) indicates mitogenomes created in this study

Because all phylogenetic analyses recovered the same protein-coding gene topology, we used the tree file generated from the gene- and codon-partitioned dataset for subsequent codeml analyses. Table 4 illustrates ω values and comparative results for the branch and site models. The free-ratios branch model outperformed the null (one-site, M1) and two-site models, indicating differential selection among all lineages. The ω values produced by this model suggest that all lineages except At. belzebuth are evolving under negative—or purifying—selection in their mitogenomic PCGs (ω < 1). The likelihood ratios test found that the site model allowing heterogenous ω over sites (M2a) provides a better fit than branch model M0, implying that selection pressure varies across codons. However, models permitting sites with ω > 1 (M2s and M8) did not provide a better fit than their nested null counterparts, indicating no codons under clear positive selection. Further, while the BEB calculation found 114 positively selected sites, the probability of these sites actually having ω > 1 was never over 75% (Table S6).

Table 4.

Calculated ω values for each branch and site models, with their comparative significance. P values are calculated through a chi-squared test of the likelihood ratios test statistic (2ΔlnL)

Model set	Model	lnL	ω	2ΔlnL	p value
Branch models	One Ratio (M0)	-36266.37	0. 641
	Free Ratio (M1)	-36261.19	Ao. azarai 0.688	M1 vs. M0 = 10.34	0.0012
			Al. caraya 0.604
			Al. palliata 0.609
			B. arachnoides 0.549
			L. l. poeppigii 0.678
			L. flavicauda 0.658
			At. chamek 0.947
			At. belzebuth 0. 451
	Two Ratio (M21)	-36266.33	ω1 = 0.696	M2 vs. M0 = 0.067	0.79
	(L. flavicauda as foreground)		ω2  = 0.666
	Two Ratio (M22)	-36266.11	ω1 = 0.646	M22 vs. M21 = 0.442	0.506
	(Lagothrix spp. as foreground)		ω2  = 0.593
Site Models	M1a	-36083.74	0.641	M1a vs. M0 = 365.25	2.02E-81
	M2a	-36083.74	0.682	M2a vs. M1a = 0	0.99
	M7	-36084.73	0.672	M7 vs. M8 = 1.99	0.36
	M8	-36083.74	0.678

ω_{1 =} background ω for branch-site models

ω₂ = foreground ω for branch-site models

Under the branch-site Model A, the BEB Analysis did not find any sites to be under positive selection in the foreground (L. flavicauda) branch. Likewise, the branch-site Model A did not provide a better fit than the null branch-site model (p = 0.99). It did, however, suggest that roughly one third of sites (32%) showed evidence of purifying selection rather than neutral evolution (68%) (Table 5).

Table 5.

Estimates of proportions of sites within each site class in the ω distribution under the branch-site Model A. The different classes are defined as follows: Sites from site class 0 are under purifying selection along all branches with 0 < ω < 1, while all branches in site class 1 are under neutral evolution (ω = 1). Sites in classes 2a and b are under positive selection along foreground branches (ω₂ > 1), while the background branches are under purifying selection (0 < ω₁ < 1) or neutral evolution (ω₁ = 1)

Site Class	Proportion	Background ω	Foreground ω
0	0.3204	0.0068	0.0068
1	0.6796	1	1
2a	0	0.0068	1
2b	0	1	1

Divergence time estimation

Figure 4 illustrates the divergence time estimates from the BEAST2 analysis. The posterior probabilities for each node on the tree inferring divergence dating estimates were at or close to one. Using median node heights as timescale estimates within our clade of interest (Atelinae), the genera Lagothrix and Brachyteles diverged ~ 11.1 million years ago (Ma) (95% highest posterior density interval [HPD]: 7.52, 15.24), and this clade in turn diverged from a common ancestor with Ateles ~ 13.8 (10.26, 18.19) Ma. In this analysis, Lagothrix spp. share a common ancestor at 4.98 (2.50, 8.20) Ma, which is older than the MRCA of the Ateles species used in this study, which was dated at 2.83 (1.07, 6.29) Ma.

Fig. 4.

Phylogenetic tree with primate divergence time estimates inferred from 4 combined BEAST2 runs at 25 million generations each. Node bars represent 95% highest posterior density range around the median divergence date estimate (indicated in node labels). The scale axis represents millions of years ago and asterisks (*) by species names indicate mitogenomes created in this study

Discussion

The primary goal of this study was to investigate whether long-term occupation of Andean montane forests has left detectable signatures of selection in mitochondrial genes associated with energy production via OXPHOS pathways. Contrary to our predictions, we found no evidence of positive selection acting on mitochondrial protein-coding regions in L. flavicauda, nor did we find differential selection patterns in the broader Lagothrix lineage. Instead, results suggest pervasive purifying selection across protein-coding regions in almost all taxa included, indicating functional conservation despite substantial ecological differences between high- and low-altitude habitats.

We successfully used ONT sequencing to construct mitogenomes for two wild woolly monkey taxa in order to test our hypotheses, with one derived from low-quality fecal samples, illustrating the power of using this portable sequencing technology for fecal mitogenomics. The low quality of our 15-year-old fecal samples with irregular cold chain maintenance did, however, require PCR amplification to sequence the full L. flavicauda mitogenome. It is notable that we initially attempted to acquire mitochondrial DNA through shotgun sequencing, and while this resulted in a similar number of sequencing reads, the large majority were of exogenous DNA (namely bacterial, regardless of attempts at host DNA capture using methods such as those by [109]), resulting in too few mitochondrial reads to cover the mitogenome. This contrasts with the reference we constructed for L. lagotricha poeppigii, which we were able to complete without PCR from gDNA extracted from fresh tissue. Our reliance on PCR and adaptive sequencing to amplify the mitogenomic content from fecal DNA explains both the relatively high percentage of L. flavicauda reads mapping to the reference mitogenome, and the extraordinarily high depth of coverage (> 9,000X) of our reference. While our attempt to sequence a viable reference from multiple low-quality fecal samples ultimately necessitated PCR to get sufficient mitogenomic reads for assembly, Wanner et al. [66] were able to obtain a 258X mitogenome from another platyrrhine primate using a single, freshly collected fecal sample without any enrichment aside from adaptive sampling, demonstrating the versatility of the ONT platform for mitogenomic sequencing across a broad range of fecal sample quality. Together, these sequences exemplify the utility of ONT sequencing for low-quality samples and—in the case of the L. lagotricha poeppigii sample that was sequenced at TBS—successful in situ data generation, and adds to the growing research utilizing portable ONT sequencing in conservation genomics of endangered non-model species [110–115].

As noted in other published primate mitogenomes, the organization of the mitogenome for these two taxa is highly conserved [66], with the coding regions nearly identical to those in a previously published L. lagotricha mitogenome (NC_021951). Our two newly constructed references were very similar in size with the L. flavicauda reference at 16,603 bp long and the L. lagotricha poeppigii reference at 16,590 bp long. Additionally, comparing our L. flavicauda reference to the unpublished sequence by Di Fiore et al. [24] showed just 19 bp differences. However, the Di Fiore et al. sequence had a significantly shorter D-loop region, by almost 400 bp. Considering the D-loop length of our reference aligned well with that of other ateline species, we consider this to be a large gap in the Sanger-produced reference, further illustrating the power of a high-coverage long-read ONT reference.

Both our ML and Bayesian phylogenetic analyses yielded currently accepted relationships between atelid taxa, all with high node support. This topology corroborates other phylogenetic studies including the Atelidae family that use mitochondrial DNA [24], chromosomal data [116] and nuclear DNA [116]; although, some nuclear phylogenies place Ateles and Brachyteles as sister clades, rather than Lagothrix and Brachyteles [117]. We also found similar divergence patterns among atelids as previous studies, and provide further support for a relatively early divergence of taxa within the genus Lagothrix compared to other ateline genera [52, 71, 118]; however, the relatively older divergence date found here may be due to the inclusion of older node calibrations, which tend to be more uncertain than more recent calibrations [107, 116]. Additionally, because taxon sampling was designed primarily to resolve relationships within Lagothrix and other atelids rather than to comprehensively represent primate diversity, divergence estimates should be interpreted primarily as a temporal framework for relative divergences within the focal clade rather than precise estimates of primate-wide divergence times.

Our current phylogenetic analysis clearly placed L. flavicauda outside of the clade including L. lagotricha subspecies; however, a lack of monophyly between L. lagotricha subspecies has been previously illustrated through population genetic analyses [50], though this could be better supported by population-level data from all species. We estimated a median divergence date of 2.8 Ma for L. lagotricha and our L. lagotricha poeppigii reference, which aligns with late Pliocene uplift and reorganization of Andean and Amazonian landscapes that altered habitat connectivity and likely increased opportunities for isolation between montane and lowland populations [119]. Although our data cannot distinguish whether montane adaptations evolved specifically within the L. flavicauda lineage or were inherited from an ancestrally montane Lagothrix, the divergence estimate places lineage diversification within a timeframe when both Andean uplift and climatic change could have contributed to ecological differentiation between montane and lowland woolly monkeys. Additionally, the split between these species predates many other platyrrhine species-level distinctions, raising the long-debated question about whether L. lagotricha subspecies should be elevated to species ranks. However, despite the relatively deep divergence time inferred here, previous work suggests caution in the elevation of woolly monkey subspecies to species rank, as genetic distances have been reported to be substantially lower between L. lagotricha subspecies than typically recognized platyrrhine species [50]. Application of genetic and phylogenetic species concepts to woolly monkeys has therefore produced mixed results. As our divergence estimates are based on mitochondrial genomes from only two L. lagotricha subspecies, any classification is beyond the scope of this study. Importantly, further unpacking of the pattern and timing of divergence within the Lagothrix clade, especially to understand the evolutionary history in relation to altitude change in L. flavicauda relative to the lowland woolly monkeys, requires expanded mitogenomic sampling efforts from this species and the subspecies subsumed within L. lagotricha.

We evaluated two alternative hypotheses regarding the evolutionary history of montane adaptation in Lagothrix. Under the first hypothesis, we predicted that L. flavicauda would show signs of positive selection in protein-coding regions, indicating that it evolved following its restriction to high-elevation forests. Our results did not support this prediction, as we did not detect any genes or loci undergoing positive selection in the mitogenome of L. flavicauda as a potential adaptative response to exclusively occupying high elevation yungas. Further, our alternative hypothesis predicted that adaptations to montane forests arose earlier in the evolutionary history of Lagothrix due to the proposal that ancestral Lagothrix occupied Andean environments before the expansion of some lineages into Amazonian lowlands [38], in which case, evidence of selection would be detectable across the broader Lagothrix lineage relative to other atelids. While branch models did show slight purifying selection in the Lagothrix lineage, it was not different than background lineages. Against either of these predictions, our free-ratios model, which was the best fit branch model of the analysis, and the site models indicate that the protein-coding regions of the mitogenomes of all ateline lineages, including L. flavicauda, appear to have experienced purifying selection. Comparative studies of humans living in high- and low-elevation environments have also shown associated differences in these gene regions [120, 121], but the functional importance of these differences and how they might mitigate the physiological challenges of mild hypoxia and colder temperatures at this elevation (~ 1800 masl) are unclear. Overall, purifying selection (ω < 1) in the branch models indicate the functional conservation of genes important to the OXPHOS pathway in L. flavicauda. While d_N/d_S analyses commonly use one representative sequence per species (e.g. [55, 57]), population-level sampling is necessary to distinguish lineage-wide evolutionary patterns from individual variation and to evaluate whether genetic variants associated with high-elevation occupation are consistently maintained across populations.

The absence of detectable selection in mitochondrial protein-coding regions should not be interpreted as evidence that L. flavicauda lacks adaptations to high-altitude environments for several reasons. For one, while the mitogenome is a marker well-suited to approach evolutionary questions, its maternal inheritance and representation of a single locus means that it may not reflect evolutionary patterns presented by the rest of the genome. Further, these habitats impose multiple selective pressures, including hypoxic conditions, reduced temperatures and reduced resource availability, and adaptations to these conditions are not mutually exclusive. Our results suggest that genetic responses to these challenges may not be reflected in the mitochondrial genome, specifically. For example, L. flavicauda has longer and denser fur than its congeners [29], consistent with adaptation to cooler montane climates rather than low tropical forests. Additionally, montane tropical environments impose narrow thermal niches on taxa, where tolerance to relatively cool and stable cloud forest conditions may play a strong role in defining elevational range [9, 10]. This has been exemplified by comparative work on temperate versus tropical birds, where tropical montane taxa have narrower climatic ranges and limited dispersal outside of those ranges [122]. The genetic basis for high-elevation adaptation should be further explored through whole-genome sequencing or by utilizing other genetic biomarkers. For instance, variation in EPAS1—including exonic variants, regulatory variation, and methylation—have characterized adaptive physiological responses (e.g. lower hemoglobin, mitigated pulmonary construction, and other vascular responses) to high altitudes in human populations [122, 123]. This highlights the need for whole-genome sequencing for L. flavicauda to more comprehensively investigate the genetic basis for its adaptation to high elevations.

This study provides the first publicly available high-coverage mitogenome for L. flavicauda and establishes a foundation for investigating the evolutionary consequences for long-term occupation of high-elevation environments for this platyrrhine species. Although we found purifying selection in L. flavicauda that did not differ from congeners or other atelid species, these results suggest their evolution in the Andes may involve alternative genetic pathways or mechanisms not captured by mitochondrial sequence evolution. Given its Critically Endangered status, more research is required to understand the biology of L. flavicauda and how it has evolved to occupy its current niche.

Abbreviations

ONT

Oxford Nanopore Technologies

OXPHOS

Oxidative phosphorylation

mtDNA

Mitochondrial DNA

dN/dS and ω

Nonsynonymous to synonymous mutation ratio

Bp

Basepairs

masl

Meters above sea level

Spp.

Species (plural)

NGO

Nongovernmental organization

TBS

Tiputini Biodiversity Station

CITES

The Convention on International Trade in Endangered Species of Wild Flora and Fauna

IACUC

Institutional Animal Care and Use Committee

PCR

Polymeraser chain reaction

SUP

Super accuracy (basecalling)

SNP

Single nucleotide polymorphism

PCGs

Protein coding genes

ML

Maximum Likelihood

AICc

Akaike Information Criterion corrected for small sample sizes

MCMC

Markov chain Monte Carlo

BF

Bayes Factor

LRT

likelihood ratio tests

BEB

Bayes Empirical Bayes

MRCA

Most recent common ancestor

CpG

Cytosine-phosphate-guanine

HPD

highest posterior density interval

Ma

Million years ago

Authors’ contributions

M.A.Z. conceptualized and designed the study, processed L. flavicauda fecal samples (extraction through sequencing), and conducted bioinformatic analyses, interpreted the data, and wrote the original draft. C.A.S. provided supervision and methodology development, acquired funding and collected L. flavicauda samples. A.D. provided supervision and funding and collected and processed the L. lagotricha poeppigii sample. S.S. and F.M.C. provided supervision and project administration. All Authors reviewed the manuscript.

Funding

This research was made possible by the National Science Foundation Graduate Fellowship Program (award number 2234657) for supporting MZ while processing L. flavicauda fecal samples and conducting bioinformatic analyses. Lab research funding by Boston University, New York University, and the University of Texas at Austin supported sample collection and the wet bench work processing the L. lagotricha poeppigii sample while in the field.

Data availability

The mitogenome sequences generated in this study have been deposited into NCBI GenBank with accession numbers PZ228854 (L. flavicauda) and PZ239894 (L. lagotricha poeppigii). Raw sequence data (FASTQ files) are available in the NCBI Sequence Reads Archive (SRA) under BioProject PRJNA1445992 and accession numbers SAMN56810639- SAMN56810640 (L. flavicauda) and SAMN56810459 (L. lagotricha poeppigii). Code associated with the analyses in this study will be published in Dryad upon publication of this manuscript.

Declarations

Ethics approval and consent to participate

All collections and research adhered to the legal requirements of the countries from which the samples were collected (Ecuador and Perú) and were carried out in compliance with The Convention on International Trade in Endangered Species of Wild Flora and Fauna (CITES). Sample collection for L. flavicauda was conducted under Peruvian Ministerio de Agricultura and Instituto Nacional de Recursos Naturales (INRENA) collection and research authorization N° 130- 2007-INRENA-IFFS-DCB issued to co-authors Sam Shanee and Fanny Cornejo. At the time that this sample was collected, NYU did not require IACUC approval for noninvasive sampling. Sample collection for L. lagotricha poeppigii tissue sample was conducted under Ecuadorian Contrato Marco number MAE-DNB-CM-2019-0117 issued to co-author Anthony Di Fiore and approved by The University of Texas at Austin IACUC (protocol number AUP-2020-00229). All animal handling and interactions are consistent with the ethical recommendations of the American Society of Primatologists.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.