SNP discovery in genomic libraries of the three studied Acrocomia species

Sequencing of the two ddGBS libraries for A. aculeata and A. totai generated a total of 219,264,253 reads, while the two A. intumescens ddGBS libraries generated a total of 554,630,990 reads. After quality-control filtering, the number of retained reads were: 60,786,924 (mean = 779,319.5 reads per sample, SD ± 449,982) for A. aculeata; 29,283,574 (mean = 697,227.9 reads per sample, SD ± 355,813) for A. totai; and 276,741,447 (mean = 1,990,945.6 reads per sample, SD ± 1,333,058) for A. intumescens. After the analysis in Stacks, the final data set had 1997 SNPs for the 78 A. aculeata samples (mean depth per locus = 18.7X, SD ± 6.2, 8.8% of missing data). For the 40 A. totai samples, 1629 SNPs were identified (mean depth per locus = 17.2X, SD ± 6.4, 2.9% of missing data). For the 131 A. intumescens samples a total of 3466 SNPs were identified (mean depth per locus = 11.5X, SD ± 4.4, 13.4% of missing data).

Putative signatures of selection and population structuration in A. aculeata, A. totai, and A. intumescens

Because pcadapt and LFMM are methods that account for the global genetic structure of the data, some interesting patterns may be considered for the analyses of putative signatures of selection. Both analyses suggested that the genetic structure within A. aculeata and A. intumescens is not explained by the local biome groups assumed according to the classification proposed by Freitas et al. [12], but rather the outcome of genomic isolation of biogeographic barriers and historical events. In the case of A. totai, a genetic structuring was observed in accordance with local biomes and macro ecoregions (Fig 3, S1 and S3 Figs) suggesting local ecosystem adaptation. Similar DAPC results were obtained for A. totai, indicating a significant degree of divergence between natural populations across specific biomes. The major genetic structure in A. aculeata is observed among sampling locations (or biogeographic groups) within and outside Brazilian lowlands (<1500 masl), resulting in the separation of the samples into two clusters along the first principal component (43.88% of the variance), with a distinct divergence between samples from Mexico and Central America and those from Brazil (Fig 3 and S1 Fig). In A. totai, samples from Pantanal (PAN) were the most divergent, the Argentinean-Paraguayan Chaco and Espinal (CHE) samples tend to cluster in the same genetic group, while samples from the Brazilian Cerrado (CER) were more admixed than those from the other groups (Fig 3 and S1 Fig). The major genetic divergence in A. intumescens was observed between the Brazilian state of Ceará, indicating interrupted gene flow likely due to genetic isolation or a biogeographic barrier, rather than between biomes (Fig 3 and S1 Fig).

Download:

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

Fig 3. Map of sampling locations for Acrocomia aculeata (blue), A. totai (green), and A. intumescens (red), along with major genetic divergences among local biome groups as suggested by sparse non-negative matrix factorization (sNMF) analyses.

Pie charts show the average sNMF ancestry coefficients across the genetic clusters represented by different color shades. Some pie charts were slightly moved to ease visualization. Acronyms follow Table 2 local biomes: CMF Central American Moist Forest; LLA Llanos; CDF Caribbean Dry Forest; ACF Atlantic Coastal Forest; CAA Caatinga; AMA Amazon; CER Cerrado; PAN Pantanal; CHE Chaco and Espinal. Shapefile of limits of countries in South America from IBGE-Mapas (IBGE—Brazilian Institute of Geography and Statistics; Available at: https://geoftp.ibge.gov.br/cartas_e_mapas/bases_cartograficas_continuas/bc250/versao2021/shapefile/bc250_shapefile_2021_11_18.zip) The map was developed by the authors using QGIS Version 3.32.3 (Available at: < https://qgis.org/en/site/>). DATUM: SIRGAS 2000.

https://doi.org/10.1371/journal.pone.0324340.g003

The numbers of outlier SNPs detected for A. aculeata were 500 based on pcadapt and 129 based on FstHet, while LFMM indicated 525 loci associated with at least one bioclimatic variable. Of these, 326 markers were considered as putative under selection because they were indicated by at least two of these methods (Fig 4A). A total of 67 A. aculeata sequences with an outlier marker had at least one blastx hit, and 54 had at least one GO annotation. These GO terms could be summarized in 33 classes (Fig 5A). For A. totai, pcadapt indicated 218 outlier SNPs, FstHet 306, and 446 SNPs were associated to a bioclimatic variable in LFMM. Of these, 192 SNPs were considered as putative under selection (Fig 4B). A total of 38 A. totai sequences with an outlier marker had at least one blastx hit, and 33 had at least one GO annotation. These GO terms could be summarized in 28 classes (Fig 5B). For A. intumescens, pcadapt indicated 688 outlier SNPs, FstHet 157, and 1024 markers were associated with a bioclimatic variable in LFMM. Of these, 104 SNPs were considered as markers putatively under selection (Fig 4C). Only eight A. intumescens sequences with an outlier marker had at least one blastx hit, and two had GO annotations. These GO terms could be summarized in seven classes (Fig 5C). For the three species, the most common GO annotations were related with the biological processes of “metabolic process” and “cellular process”, and the molecular functions of “binding” and “catalytic activity”. GO annotations related to “response to stimulus” and “regulation” were also frequent for the proteins with similarities to the sequences with outlier SNPs of A. aculeata and A. totai.

Download:

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

Fig 4. Venn diagrams showing the number of loci identified as outliers with different methods.

The outlier SNPs detected in Pcadapt are based on the genetic structure of the PCA, while in FstHet, they are based on the betahat statistic. In LFMM, the outliers are associated with bioclimatic variables. A) Acrocomia aculeata, B) A. totai, and C) A.intumescens.

https://doi.org/10.1371/journal.pone.0324340.g004

Download:

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

Fig 5. Summary of Gene Ontology (GO) annotations associated with proteins.

The outlier SNPs in gene sequences associated with proteins are deposited in GenBank. GO terms summarize annotations according to cellular components, molecular functions, and biological processes. A) Acrocomia aculeata, B) A. totai, and C) A.intumescens.

https://doi.org/10.1371/journal.pone.0324340.g005

These genes exhibited high similarity to genes annotated and described in palm species such as Elaeis guineensis, Phoenix dactylifera L., and Cocos nucifera L., all of which have fully sequenced genomes (Table 1, S2 Table). The genes observed were related to many functions, including oil production metabolism/catabolism, carotenoid biosynthesis, plant growth and development, organ sizes, root, flowering, pathogen resistance, and biotic and abiotic stresses. The most relevant loci identified as outliers in different methods and for each species are listed in Table 1 (the whole list is in S1 Table), while a summary of blastx results and their associated GO annotations are in S2 Table.

Download:

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

Table 1. Most relevant gene loci with putative selective signatures in Acrocomia aculeata, A. totai, and A. intumescens. Gene loci are divided into four major biological functional groups: (1) fatty acid and triacylglycerol biosynthesis, (2) carotenoid biosynthesis and plant development, (3) plant defense and disease resistance, and (4) abiotic and environmental stress. The complete list of gene loci and their description can be found in S2 Table.

https://doi.org/10.1371/journal.pone.0324340.t001

Our comparative analysis across the three Acrocomia species (A. aculeata, A. totai, and A. intumescens) revealed both common and species-specific patterns in their adaptation mechanisms to environmental stressors. While all species showed significant evidence of selection in response to bioclimatic variables associated to annual temperature (BIO 1), diurnal range (BIO 2), annual precipitation and especially in the driest month (BIO 12, 14), the degree of genetic divergence and the nature of the putatively selected loci varied notably. A. aculeata exhibited a strong genetic structure linked to biogeographic barriers, with outlier loci primarily associated with abiotic stress responses, such as fatty acid biosynthesis and carotenoid metabolism, suggesting a robust adaptation to fluctuating environmental conditions. In contrast, A. totai showed more localized genetic structuring in relation to specific biomes, with genes related to lipid metabolism, plant defense, and abiotic stress tolerance playing key roles in its adaptation to different ecological zones, such as the Pantanal and Cerrado. A. intumescens, on the other hand, displayed a notable divergence primarily in samples from Ceará, indicating possible genetic isolation and local adaptation to unique environmental pressures. Despite these differences, all three species shared common functional pathways related to metabolic processes, cellular functions, and stress responses, such as plant defense and abiotic stress tolerance. These findings highlight the diverse adaptive strategies employed by Acrocomia species, emphasizing the complex interplay between genetic isolation, local environmental factors, and evolutionary pressures in shaping their genomic diversity.

The adaptation and evolutionary process in Acrocomia: dispersal and biogeographic barriers

Acrocomia phenotypes have been shaped by ecological interactions that have influenced the evolution of specific traits for its survival [66]. Acrocomia species are distributed across tropical Americas, extending from Central America (Northern Mexico) to South America, inhabiting areas transitioning from tropical and subtropical rainforests (moist broadleaf forests) to savannah and xeric shrubland regions, including dry broadleaf forests (semi-arid and arid ecosystems) (Fig 2). Their wide distribution suggests their ability to adapt to various soil and climate conditions. sNMF analyses can help identify genetic signatures associated with adaptation to specific environmental conditions or phenotypic traits. Our findings provide insights into the genetic structure and the association of specific SNP loci under selection forces in the adaptation process of Acrocomia aculeata, A. totai, and A. intumescens (Fig 4, Table 1).

Our results support the division of A. aculeata into two major gene pools: a Central American group and a South American group, consistent with previous observations based only on neutral loci [3] (Figs 2 and 3). Notably, we found no significant differences in genetic structure within minor local biomes within each gene pool, indicating that the Amazon basin likely served as a source region for diversification across the species distribution. This pattern in genetic structure is also present in other crop species such as common beans and maize [67,68]. Both dispersal and biogeographic barriers likely influenced the genomic structure of Acrocomia aculeata. Also, A. aculeata exhibited the greatest number of putative selective signatures (Table 1; S1 and S2 Tables) in genes associated with fatty acid and triacylglycerol biosynthesis, carotenoid biosynthesis, pathogen resistance and defense, as well as genes specialized in adapting to abiotic and environmental stress. From the nine species of the genus, A. aculeata exhibits the highest fruit pulp oil content (65% w/w dry basis) [4].

Individual plant fitness is influenced by animal–plant interactions, encompassing mutualisms such as seed dispersal and antagonistic relationships like herbivory and seed predation [66]. For example, the morphology of Acrocomia aculeata fruits, including its size, color, and content, appears to have undergone a correlated evolution as predicted by the seed dispersal syndrome hypothesis before the domestication process by humans began in recent times (~10,000 BP) [1,19]. Additionally, spines and other mechanical defense features are common mechanisms employed by plants to avoid herbivory. Numerous studies on palms indicate that spinescence reduces herbivory by large mammals [69]. The same herbivory and seed dispersal pattern is present in A. totai and A. intumescens.

Frugivorous mammals have a preference for consuming dull-colored fruits, such as those that are green, yellow, brown, or pale orange [66,70] (Fig 1). It is hypothesized that the coccoid type fruit (drupe) of Acrocomia ancestor species may have been consumed by large herbivores in the past, with migration events like the Great American Biotic Interchange (GABI; 2.5 million BP) likely influencing its distribution through seed dispersal [70]. Consequently, certain observable fruit traits and defenses against herbivory, such as spiny trunks and hard epicarp, could be considered anachronic, representing interactions that occurred in the past [70]. The term “anachronic” denotes traits or features that are outdated or no longer relevant in the current context but may have been adaptive or functional in the past [66,70]. While most large mammals are now extinct in the Neotropics following the Late Quaternary Extinction episode, neotropical palms like Acrocomia spp. diversified in an ecological environment where mammalian assemblages were abundant in large-bodied species, approximately from 2.6 million to 23,000 BP [66,71,72]. However, our findings suggest only a certain level of connectivity between populations within the two major gene pools. Currently, cattle and humans are regarded as the primary contributors to seed dispersal for this species, highlighting its recent close association with human activities [69,73].

Archaeological evidence encompassing Pre-Columbian civilizations across the neotropics, from Mesoamerica (northern Mexico, Central America) to South America, suggests the cultivation and ritualistic utilization of Acrocomia aculeata [18,19]. The presence of Acrocomia in these regions indicates a potential influence of Pre-columbian migration on species diversification and offers insights into its domestication process. Such historical records reinforce the relationship between indigenous populations and Acrocomia, potentially shaping its genetic diversity and evolutionary trajectory through intentional cultivation and cultural practices. Further investigation is necessary to elucidate the specific historical events and ecological factors shaping the observed genetic structure and center of origin.

Acrocomia totai exhibits a pronounced genetic structuring, showing a clear differentiation between populations from the Brazilian Cerrado, Pantanal, and Chaco and Espinal (Figs 2 and 3). This contrasts with previous studies focused only on neutral loci using SNPs and microsatellites where they observed less genetic structuring among populations [3,74]. This indicates that the populations of A. totai are experiencing a recent diversification and ongoing speciation processes. A. totai has a subtropical distribution and is adapted to temperate conditions, with selective signatures found in genes associated with fatty acid and triacylglycerol biosynthesis, pathogen resistance and defense, as well as genes specialized in adapting to abiotic and environmental stress (Table 1; S1 and S2 Tables). Pantanal populations exhibit lower levels of admixture compared to other genomic groups within A. totai, particularly those distributed in the Brazilian Cerrado. The Cerrado lowlands and ecotones with other local biomes serve as a bridge, including for the South American gene pool of A. aculeata. Hybridization between these two species has been reported [3]. Phenotypic differences were also observed in other studies, compared to fruits from the Pantanal, fruits from Cerrado had a higher proportion of epicarp and a lower proportion of kernels [5]. The Pantanal biome, an active sedimentary basin characterized by faults and earthquakes, experiences subsidence and depressions that are highly susceptible to flooding. During the late Pleistocene (14,000–10,000 BP), arid conditions predominated, possibly allowing A. totai to recently colonize the biome [75]. The Pantanal lacks endemic tree species, with the majority of terrestrial species being immigrants from the Cerrado [76]. Similar biogeographic patterns are observed in the palm Copernicia alba Morong [77]. The domestication process of A. totai remains poorly understood, although archaeological records indicate its use for fiber production in Argentina [19]. Human utilization varies among Acrocomia species, with A. totai being utilized for the yellowish pulp of the fruit for flour production, and the oil derived from the seed, which are the two products of greatest interest [5,13,78].

A. intumescens presented a genetic structure influenced by an allopatric isolation across various phytophysiognomies within Caatinga, rather than by variations in biomes (Figs 2 and 3) [79,80]. Specifically, notable genetic structuring was observed between populations in the North Sertaneja depression and those transitioning from the Borborema highlands to the South Sertaneja depression within the Caatinga biome. The Caatinga exhibits high levels of endemism, with species adapted to survive in its arid climate characterized by drought conditions. The long-term stability of the Caatinga, along with the assembly of ancient plants, has been significantly influenced by aridification processes, while recent vegetation shifts and climate change have driven in situ diversifications. The increased environmental variability has led to the appearance of modern species through Pleistocene/Pliocene (2.6 million BP) ecological specialization [81]. This could elucidate the phylogenomic proximity observed between A. aculeata and A. intumescens, suggesting a shared common ancestor in the past and probably, Acrocomia recently colonized and diversified in the Caatinga [82]. This shared ancestry may be attributed to the dominance of tropical rainforests in the region, which connected the area to the Amazon during the Early Cenozoic period [3,83]. This pattern of endemism and genetic differentiation is shared with other plant species in the Caatinga, including other palms such as Copernicia prunifera (Miller) H.E. Moore, legumes (Coursetia, Vatairea, and Luetzelburgia genera), as well as other clades such as Conopophaga cearae (Aves. common name: Caatinga gnateater) and lizard species [80,81,84,85]. In A. intumescens, the only selective signature found in a gene was associated with stress signaling.

The close relationships observed among Acrocomia species may be attributed to the radiation of the Bactridinae subtribe towards the end of the Eocene (23 million BP), when the ancestor of Acrocomia diversified [86]. This period of divergence coincided with the terminal Eocene cooling event, characterized by substantial climatic changes that led to accelerated turnover in flora and fauna [87]. This pattern is also evident in the conserved genomic structures and similarity of Acrocomia aculeata, A. totai and, A. intumescens in their plastomes architecture and nuclear gene phylogeny, low pairwise Fst indicating a close relationships [3,83]. The high admixture between populations and the short branch lengths observed in phylogenomic studies, which separate the three species, presents challenges in establishing unambiguous intergeneric relationships within the genus Acrocomia. This may suggest a large-scale species-level extinction followed by rapid diversification of surviving lineages.

Correlations between allele frequencies and climatic variables in the LFMM analysis revealed five climatic variables influencing adaptation in the three species: annual mean temperature (BIO1), mean diurnal range (BIO2), isothermality (BIO3), annual precipitation (BIO12), and precipitation of the driest month (BIO14), indicating selection as a driving force of evolutionary adaptation (S2 Fig). Similar patterns were observed in photosynthesis rates and fruit pulp mass development in other Acrocomia studies and in the neotropical palm Mauritia flexuosa Mart., which is influenced by BIO1, BIO2, and BIO14 [7,88]. Although Acrocomia can thrive in a range of climatic conditions from humid to semi-arid ecosystems, they are found exclusively in open areas that are often influenced by human activities. Our results are supported by previous studies that suggest temperature played a significant role in the evolution of the climate niche in Neotropical palm species [9,89].

The sNMF analysis (S2 Fig) showed low differentiation among populations from different biomes in putative SNPs under selection. However, there was evidence of admixture in Acrocomia totai, particularly between natural populations from Cerrado in Paraguay, Mato Grosso do Sul (Brazil), Parana (Brazil), and Sao Paulo (Brazil). Biogeographic isolation by barriers is an important driver of genetic differentiation in palm as described by Eiserhardt et al.: “The significant effect of spatial distance on clade turnover in American palms can thus be interpreted as evidence for dispersal limitation on evolutionary timescales caused either by barriers or by time” [9]. Even though Eiserharth's results focus on macroevolutionary diversification and phylogenetic turnover by dispersal and niche evolution, they indicate that barrier distance explains more variation in diversity than spatial distance. We observe a similar pattern in our results, where biogeographic barriers are more relevant in shaping genetic structure than biomes proximity and microclimatic variables in Acrocomia niches (Fig 3).

Palms are not naturally adapted to withstand extremely high or low temperatures. Interestingly, the evolution of these geographic range restrictions does not appear to be a limiting factor for Acrocomia, despite the fact that water supply most likely limits the regional ranges of many neotropical palms [90]. Acrocomia appears to be highly adapted to semi-arid ecosystems. The distributions of neotropical palm species are assumed to be impacted by edaphic environments, despite the absence of empirical data. It is known that the natural occurrence in Brazilian populations of A. aculeata is associated with eutrophic (high fertility) soils with medium to clayey textures and an average pH of 5.5 [91]. However, further investigation is needed for the Central American gene pool characteristics.

Domestication syndrome in palms is poorly understood [16,92]. Detecting selective sweep regions associated with domestication in Acrocomia is crucial. These genes probably have undergone strong positive selection during this process. Selective sweeps are characterized by a reduction in genetic diversity around the selected allele due to the rapid increase in frequency of advantageous alleles favored by human-driven selection [93]. Maybe there is a heterogeneous origin in the domestication of Acrocomia. However, the existence of several founder lineages and the link between a mosaic ancestry patterns only will be clear with a sequenced reference genome for the three species. The identification of loci under selection can be greatly enhanced by having access to a reference genome [21]. Acrocomia may respond polygenically to environmental variation like other palms [25,88,94], meaning that selection may result in minimal changes in allele frequencies, leading to a poor adaptation signal at a single locus.

Our results demonstrate that genetic structure in populations from different biomes is homogenized within the same biogeographic regions due to gene flow. This also suggests the potential for migration to disperse advantageous alleles among populations sharing the same biogeographic region, leading to allele fixation in genes involved in adaptation to environmental change. This potentially explains Acrocomia's tolerance to seasonality and sheds light on the species’ interaction with humans and its domestication process. We suggest investigating the functionality of distinct biological pathways under diverse conditions using a transcriptomic approach for understanding better its domestication syndrome and its potential resilience to climate change [7,89]. Also, conservation strategies can be formulated by identifying genetically distinct populations that may require special attention for conservation.

Genes putatively involved in adaptation and implications for domestication in Acrocomia

A high-quality reference genome sequence and annotation at chromosomal level of plant species is essential for genetic research on crop breeding and domestication [20]. Acrocomia is considered an incipiently domesticated plant and does not have a reference genome yet, so a population genomics approach using GBS or RAD-Seq can help investigate the evolutionary mechanisms underlying its diversification and variation. Significant genomic alterations have occurred during plant domestication as a result of evolutionary processes such as genetic drift and artificial selection. Studies of artificial selection use genomic techniques to identify signatures that are likely to be subject to selection, with a focus on identifying polymorphisms in trait loci that have been masked by neutral variation in species undergoing a domestication process [21].

Despite lacking a complete reference genome for Acrocomia, we were able to analyze genes potentially under selection. To do this, Blast2GO and Blastx use different reference genomes of related palm species to compare the loci where signatures of potential adaptation were located. The most common similarity matches between our GBS tags and annotated sequences were found mostly in homologous genes of Elaeis guineensis (African oil palm), Cocos nucifera (coconut), and Phoenix dactylifera (date palm) (Table 1). For example, phylogenomic studies utilizing chloroplast genome sequences of A. aculeata and E. guineensis demonstrated that all nodes had a posterior probability of 1.0 (PP = 1.0), indicating a close relationship between these species, eventhough male and female inflorescences are separated in E. guineensis while both occur in the same inflorescence in Acrocomia spp. [83]. Here, we present some relevant candidate genes that exhibit selection signatures and could be involved in the adaptation processes (and maybe the domestication) of Acrocomia.

Some significant selection signatures were identified in genes associated with fatty acid and triacylglycerol biosynthesis pathways in A. aculeata (Table 1). For example, loci such as CLocus_25725 correspond to the gene encoding a lipid-binding protein that facilitates lipid transport. CLocus_11291 corresponds to the gene encoding a patatin-like protein 2, which exhibits phospholipase activity in lipid catabolic processes and acylglycerol lipase activity. Additionally, CLocus_32860 was found in the gene encoding an ABC transporter D family member 1, involved in lipid metabolism functions such as the transport of long-chain fatty acids, import of long-chain fatty acids into peroxisomes, and fatty-acyl-CoA transport. Furthermore, CLocus_27100 corresponds to the gene encoding a putative WRKY transcription factor that modulates the content of fatty acids and influences the accumulation of seed oil. Additionally, we found signatures like CLocus_24655 and CLocus_12388 related to genes involved in zinc ion binding processes. In the research led by Couto et al., candidate genes related to oil production traits in A. aculeata were identified [95]. The genome-wide association study (GWAS) suggested that candidate genes controlling oil production were associated with metal ion binding and correlated with traits such as fruit oil content, fruit pulp fresh mass, leaf number, and leaf length. In A. totai, the locus CLocus_17145 was present in the gene encoding GDSL esterase/lipase, which mobilizes the lipids stored in seeds and plays a major role in seed germination and early seedling establishment [52]. These SNPs could be of agronomic importance as one of the primary objectives of the Acrocomia breeding program is to increase oil content for industrial purposes such as biofuels.

These fatty acids and triacylglycerols biosynthesis pathways were also reported in the fine mapping and cross-validation of QTLs linked to fatty acid composition in different varieties of E. guineensis made by Ding et al. [52]. The Blast results for their QTL included genes and transcription factors linked to diacylglycerol acyltransferase (DGAT1) and long chain acyl-CoA synthetase. The synthesis of triacylglycerol and fatty acids occurs in different compartments within plants. Specifically, new fatty acid synthesis takes place in the plastid (leucoplast), where the acyl carrier protein (ACP) holds the fatty acid chain as it elongates. Then, acyl-ACP thioesterases hydrolyze the acyl-ACPs, releasing non-esterified fatty acids. These fatty acids are then exported to the endoplasmic reticulum, where they assemble to form triacylglycerol [25,96].

Another significant finding was signatures associated in carotenoid biosynthesis pathways (Table 1). In A. aculeata, loci such as CLocus_16512 and CLocus_11820 were associated with genes encoding carotenoid cleavage dioxygenase 7 and carotenoid cleavage dioxygenase 8 homolog B, which play roles in carotene catabolic processes and the production of apocarotenoids. These enzymes, known as carotenoid cleavage dioxygenases (CCDs), also influence shoot and branching inhibition and act on various carotenoid substrates like lycopene or zeaxanthin. Additionally, in A. totai, the locus CLocus_24114 was found in the gene encoding 9-cis-epoxycarotenoid dioxygenase NCED6, which is involved in regulating plant growth and seed dormancy. Carotenoids are vital for plant growth and development as they serve as precursors for the synthesis of plant hormones such as strigolactones and abscisic acid (ABA) [34,35]. From a nutritional perspective, the high content of carotenoids in crops is an attractive trait, offering benefits such as mitigating vitamin A deficiency through dietary intake of plant-derived carotenoids. For example, the increasing popularity of orange carrots (Daucus carota subsp. sativus) may be attributed, at least in part, to the numerous studies highlighting the health advantages associated with carotenoids [97].

Additionally, we observed the presence of outlier SNPs in putative resistance genes from different classes, indicating that natural populations under selection possess important genetic resources for crop defense against pathogens (Table 1). For instance, outlier SNPs such as CLocus_27100, CLocus_317, CLocus_22454, CLocus_29621, CLocus_35203, and CLocus_11291 were identified in genes involved in plant defense and disease resistance responses in A. aculeata. In A. totai, SNPs such as CLocus_37219, CLocus_29376, and CLocus_12178 were associated with genes involved in plant-pathogen recognition, immune priming, and regulation of defense response to fungus, respectively. Currently, there is local evidence of pests affecting both, natural populations and commercial plantations of A. aculeata. Examples include Cyclocephala forsteri (family Melolonthidae) and the mealybug Dysmicoccus brevipes (Hemiptera: Pseudococcidae), particularly in plantations located in Brazil's central region that are now in the production phase [4,98,99]. Phytosanitary issues with this novel crop would likely arise during the domestication bottleneck in response to genetic diversity and loss of heterozygosity.

We have identified signatures in genes specialized in adapting to abiotic and environmental stress in A. aculeata, such as CLocus_317, CLocus_12388, CLocus_23382, CLocus_25856, and CLocus_32860 (Table 1). In A. totai, a signature like CLocus_15942 and CLocus_24114 were found in the gene encoding low-affinity sulfate transporter 3, and 9-cis-epoxycarotenoid dioxygenase NCED6, respectively, responsible for the plant's response to drought and salinity stress [58]. Similarly, in A. intumescens, the only signature (CLocus_111127) present on a gene was also associated with stress responses. This may help elucidate how Acrocomia species can withstand extreme temperatures and droughts. Drought tolerance is a polygenic, complex quantitative trait [100]. Developments in agricultural physiology and genetics have contributed with important insights into drought tolerance in palm species [94,100,101]. Therefore, improving yield under water-deficient conditions is a major objective in plant breeding. Increasing drought tolerance through traditional breeding is a slow process [102]. Acrocomia's inherited characteristics give it the advantage of being adapted to semi-arid, even arid regions like the Brazilian Caatingas. For the majority of non-model species, the mechanisms driving drought tolerance remain poorly understood and vary among species. In comparison to domesticated species, incipiently or less domesticated populations may have a larger range of ecological adaptations [22]. Crop adaptability to perturbed areas depends on its resistance and resilience to environmental stress, which could be crucial for Acrocomia's adaptation in the face of climate change.

Challenges of reduced representation sequencing (GBS/RAD-seq) in crop adaptation studies

Reduced representation sequencing methods in adaptation studies such as genotyping-by-sequencing (GBS) or restriction-site associated DNA sequencing (RAD-seq) employ high-throughput sequencing to generate high genome-wide marker data. The advantages of these techniques is their capacity to develop genotyping assays without requiring prior genomic knowledge or substantial costs without needing a reference genome for perennial non-model crops like Acrocomia [20]. GBS serves as a great alternative for investigating unconventional food plants, natural populations, and wild relatives. Additionally, it has been employed to assess diversity and gene flow between crops and their wild relatives, as demonstrated in Brazilian manioc varieties and other crops [23]. Some challenges in adaptation analysis using GBS data include allelic dropout [103], missing targets in resequencing, false positives selection signatures [104] and the management of missing data [105]. However, the identification of false positives may arise from departures from the model's assumptions and covariance with sampling techniques, population structure, and demographic variables [104]. Generally, combining the SNPs identified by various outlier tests is one approach to address these limitations [23,93].

Meanwhile, it is becoming increasingly evident that many crops have complex ancestries, necessitating additional archaeogenomics data [21]. Because of its wide distribution, to gain a better understanding of the domestication process of Acrocomia and possible genes targeted by human selection, a more comprehensive examination of variables including the diversity of wild population substructures and the proportional genetic gains of various natural subpopulations is recommended, especially in the Central American genepool. A pangenomic approach is crucial to consider because plant genomes are dynamic, containing a reservoir of genetic diversity that enables adaptation to different biomes [22]. Future crop varieties will likely require a broader range of genes than those provided by single reference genomes. Developing a pangenome of Acrocomia could capture the genomic diversity present in the various gene pools observed in our research.

The promising aspects of Acrocomia and future perspectives

After exploring various aspects of population genetics, adaptation processes, and potential target genes for domestication in Acrocomia species, our study presents a comprehensive understanding of the genetic structure, evolutionary history, and adaptive mechanisms within this genus. Through advanced genomic techniques such as SNP genotyping and adaptive signatures analyses, we explored the relationships between different Acrocomia species and their adaptation to diverse environmental conditions across neotropical biomes. A high degree of diversity was observed in Acrocomia species from various neotropical biomes. This variation may have arisen from the crop's natural history, domestication process and its cultivation in diverse ecosystems with varying human preferences. It is plausible that some of the selection signals are associated with desirable agronomic traits, which could be crucial for the breeding of these palms.

Acrocomia spp. emerges as a promising crops due to their multipurpose potential, offering prospects in oil production, biofuels, food, and pharmaceutical industry [106]. Notably, the oil extracted from A. aculeata exhibits quality comparable to African palm oil (E. guineensis) while requiring lower water resources and demonstrating adaptability to semi-arid to arid environments and degraded ecosystems. Given that approximately 80% of Brazilian grasslands are classified as degraded due to anthropogenic landscape change, intercropping with perennial plants like Acrocomia emerges as a promising solution for restaurating and mitigating soil degradation [4,7]. Currently, Brazil is leading its breeding programs through public-private initiatives focusing on A. aculeata and A. totai [107]. It is projected that experimental and commercial plantations of A. aculeata will reach an area of 200,000 hectares in the next decade [108], with a guaranteed market for the oil and biodiesel production. Significant research progress has been made, particularly in seed germination, seedling production, and integrated agricultural systems, although molecular breeding remains necessary [2,4]. Prioritizing the sequencing and annotation of the entire genomes of the three most economically important species in the genus is crucial for future genomic efforts in their domestication process. Addressing challenges such as genotype-environment interaction, biodiversity prospecting for phenotypic breeding ideotypes (plant models), and crop management are necessary. Most research efforts have focused only on the South American gene pool.

Our findings reveal significant genetic structuring within and between species, highlighting the influence of dispersal, biogeographic barriers, and historical ecological interactions on the evolution of Acrocomia. We observed distinct genetic signatures associated with adaptation to specific environmental factors, including climatic variables, pathogen resistance, and stress responses. Notably, Acrocomia species exhibit potential resilience to climate change, particularly in semi-arid ecosystems. Furthermore, our study sheds light on the potential breeding of Acrocomia species, emphasizing the importance of genetic resources and candidate genes associated with desirable agronomic traits such as oil content, carotenoid biosynthesis, and stress tolerance.

These findings significantly advance the field of plant genomics and breeding by providing valuable insights into the genetic mechanisms underlying adaptation in non-model perenninal species, particularly within the context of biogeographic and environmental variation. By using advanced genomic tools and gene ontology analyses, this study provides an extensive catalog of loci potentially under selection, which can be used to inform future breeding programs focused on improving resilience and adaptation to changing environmental conditions. While facing challenges inherent to reduced representation sequencing methods, our research underscores the necessity of comprehensive genomic resources, including reference genomes and pangenomic approaches, to fully elucidate the domestication syndrome and facilitate breeding efforts.

Furthermore, the comparative analysis across species not only highlights the role of local adaptation to specific biomes but also reveals species-specific responses to environmental pressures, offering new directions for targeted genomic selection. The inclusion of non-model species in genomic studies opens up new possibilities for incorporating underexplored genetic resources into breeding programs, especially for tropical and subtropical species that are critical to agriculture and sustainable development. This approach facilitates the development of more efficient breeding strategies for crops with enhanced stress tolerance, improved productivity, and greater adaptability to climate change.

Acrocomia is a novel crop offering opportunities for future industries while contributing to sustainable development and biodiversity conservation. By prioritizing collaborative research efforts in private-public programs, genetic conservation, and policy support, different stakeholders can unlocked the potential of genetic diversity and adaptive capacity of Acrocomia species to address global challenges, including food security, renewable energy production, and climate resilience in agroecosystems.

Plant material and DNA isolation

We sampled leaves from natural populations, 78 for A. aculeata, 40 for A. totai, and 131 for A. intumescens, for a total of 249 samples (Table 2). The collections of these samples were registered in the Brazilian National Council for Genetic Patrimony CGEN (numbers A69E071 and A5D139D). The leaves were dehydrated using silica gel and stored in paper bags at −20°C. Following Doyle and Doyle's protocol, we extracted whole genomic DNA from 50 mg leaf samples. Agarose gel electrophoresis (1% w/v) with GelRed stain (Sigma-Aldrich) was used to assess DNA quality and integrity. We quantified and normalized DNA concentrations to 30 ng/μL using the dsDNA BR Assay quantification kit for the Qubit3 fluorometer (Invitrogen).

Download:

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

Table 2. Acrocomia species samples.

Geographical location and biome of the Acrocomia species samples.

https://doi.org/10.1371/journal.pone.0324340.t002

GBS libraries and SNP discovery

Genomic libraries were prepared following the protocol of genotyping-by-sequencing using two restriction enzymes (ddGBS) as described by Poland et al. [109]. The combination of NsiI and MspI (New England Biolabs) was used for the libraries of A. aculeata and A. totai, while the combination of NsiI and MseI (New England Biolabs) was used for A. intumescens. Two libraries were prepared mixing A. aculeata and A. totai samples, each using a 96-plex set of NsiI barcode adapters and a common MspI adapter. Another two libraries were prepared for A. intumescens, each using a 96-plex set of NsiI barcode adapters and a common MseI adapter. The ddGBS libraries were quantified through RT-PCR on the CFX 384 Touch Real Time PCR (BioRad, USA) equipment using a KAPA Library Quantification kit (KAPA Biosystems, USA, cat. KK4824), and the fragments’ profiles were inspected using the Agilent DNA 1000 Kit on a 2100 Bioanalyzer (Agilent Technologies, USA). The libraries of A. aculeata and A. totai were sequenced on two separate runs in an Illumina NextSeq500 (Illumina, USA), with single-end and 150 bp configurations. The two libraries of A. intumescens were sequenced on a single run, but in separate lanes, in an Illumina HiSeq3000 (Illumina, USA), with single-end and 101 bp configurations. The dataset is available in the supporting information (S1 File).

The general sequencing quality and the presence of adapters were assessed with FASTQC [110]. The SNP discovery was performed separately for each species following the de novo pipeline of the program Stacks v.1.42 [111] with similar filtering criteria. Due to the presence of adapters, sequences were trimmed to 90 bp for A. aculeata and A. totai, and 80 pb A. intumescens. Trimming, quality control (removal of sequences with uncalled bases and with Phred scores <10), and demultiplexing were performed with the module process_radtags. For each sample, loci were assembled using the module ustacks (for non-model species that lack a reference genome) considering minimum sequencing depth (-m) of 3, and distance between reads from the same loci (-N) of 2. This module groups alleles into loci using a k-mer search technique. First, a hashing method is applied to organize exactly matching reads into stacks. These stacks are then broken down into k-mers, or subsequences of length k, which are compared across stacks to identify matching alleles [111]. A catalog of loci across samples was built with the module cstacks, considering the distance between locus (-n) of 2, and loci of the samples were compared to the catalog using sstacks. The loci with lower probabilities (-lnl_lim 10) were discarded using the module rxstacks. For the three species, candidate SNPs were identified using the module populations considering a minimum depth of 3, minor allele frequency of 0.01, the presence of SNP in at least 75% of the samples in each of 14 (A. aculeata), 8 (A. totai), or 10 (A. intumescens) sampling locations. To avoid explicit linkage only a single SNP was retained per GBS locus. Quality metrics (mean sequencing depth per locus and per sample, percentage of missing data per sample) were assessed with VCFTools [112]. Additionally, samples with more than 50% of missing genotypes were removed from the final data set, resulting in 69 samples for A.aculeata, 40 for A. totai, and 131 for A. intumescens.

Detection of putative signatures of selection

Putative signatures of selection were investigated using complimentary approaches [65,113] either based on significant deviations of F_ST estimates among populations (fsthet method), based on principal component analyses (PCA) (pcadapt method), or based on environmental association analyses, which identify association between environmental variables and individual markers.

A classic approach to identify outlier SNPs is the identification of loci with extremely high or low values of F_ST (or related statistics) estimated among populations. This method was performed using the FstHet [114], package for R 3.3.1 [115] which implements a model similar to FDist2 [116] to construct a neutral envelope for the distribution of F_STs estimated based on the given data set. The neutral envelope was constructed based on 1,000 bootstraps of the F_ST-analogous betahat statistic [117], which considers variations in sampling sizes across populations. In this analysis, the markers below or above the 95% envelope of betahat estimates were considered as outlier SNPs.

Pcadapt [118] was used to identify SNP markers significantly associated with the genetic structure of the data based on a PCA, without assuming any genetic model. This analysis was performed in the pcadapt package for R 3.3.1 [115], retaining the first 3, 4, and 1 K principal components for A. aculeata, A. totai and A. intumescens, respectively (S1 Fig). In this analysis, the SNPs with q-values ≤ 0.1 (corrected p-values for their association with the first K principal components) were considered as outliers.

The Latent Factor Mixed Models (LFMM) analysis [119] was used to assess the correlations of SNP markers with environmental variables obtained in the WorldClim2 data base [120]. The information recovered from WorldClim2 refer to 19 bioclimatic variables, of which 11 are primarily related to annual trends of temperature (BIO1 to BIO11) and eight are related to annual trends of precipitation (BIO12 to BIO19). For each species, the variables were extracted for the sampling points, and a Pearson’s correlation test was employed to minimize interdependence among them, and only the variables with a correlation coefficient < 0.8 were retained (S2 Fig). A sparse non-negative matrix factorization (sNMF) analysis [121] was performed to estimate the most likely number of genetic clusters for each species to model the covariation of the subjacent genetic structure in LFMM. Both, sNMF and LFMM were performed with the package LEA [122] for R 3.3.1 [115]. For each species, 10 independent repetitions of sNMF simulating from 1 to 10 K ancestral groups were performed with 200,000 iterations. The most probable number of K = 2 for A. aculeata and A. intumescens, and K = 4 for A. totai were estimated according to the cross-entropy estimates of the algorithm (S3 Fig). These numbers were used in subsequent LFMM analyses, which were performed based on 10 repetitions of 50,000 burn-in followed by 100,000 iterations of the algorithm for each species. The generated p-values were corrected considering a false discovery rate (FDR) of 0.1 as threshold for the identification of SNPs significantly associated with the environmental variables.

SNPs were declared as putatively under selection if identified as outliers (or significantly associated to environmental variables) by at least two of the three methods described above. Blast2GO (BioBam, Spain) [123] was used to assess similarities between the GBS tags with outlier SNPs and proteins with described putative functional annotations. The similarity between GBS tags in which outlier SNPs were identified and proteins deposited in GenBank was evaluated using blastx with default configurations but restricting the search to Viridiplantae data. Sequences with blastx hits were then screened against the Pfam database to search for protein domains and their associated Gene Ontology (GO) annotations. The putative annotations were summarized based on the GO terms and visually represented with bar plots using the on-line tool WEGO (Web Gene Ontology Annotation Plotting, China) (http://wego.genomics.cn/).