Portal de Periódicos da CAPES

Opportunities for unlocking the potential of genomics for A frican trees

2015; Wiley; Volume: 210; Issue: 3 Linguagem: Inglês

10.1111/nph.13826

1469-8137

Barnabas H. Daru, Dave K. Berger, A.E. van Wyk,

Molecular Biology Techniques and Applications

Trees (or their absence) represent one of the most defining features of landscapes on the African continent. However, they face major threats including habitat loss and degradation, invasive alien species, disturbance from frequent fire, over-harvesting, pollution, changes in pollinators or dispersers populations, and climate change (Balmford et al., 2001; Davies et al., 2011). Understanding how trees respond to these impacts would require an integrative approach of which genomic science has a potentially major role to play (Plomion et al., 2016). Since the advent of genomic science, its investigative power has been exploited for trees in temperate regions, particularly involving members of Pinus, Picea, Pseudotsuga, Populus, Eucalyptus, Quercus, Castanea, Malus, Prunus, and Fraxinus (Neale & Kremer, 2011; Neale et al., 2013). These species serve as models for exploring various processes in molecular genetics, functional biology, evolutionary biology, phenotypic and genotypic adaptation, physiology and organismal development (Tuskan et al., 2006; Plomion et al., 2016). Although the tropics have exceptionally high tree diversity – with Africa alone having c. 50 times more native tree species than temperate Europe (Slik et al., 2015) – tree genomic research in this region lags behind that of temperate ones. Limited funding and the lack of reference genomes for tropical trees have limited the progress of genomic science on the African continent. With over 6000 tree species on the African continent (Slik et al., 2015), there is a need to establish reference genomes for the major tree families. First, we discuss ways to exploit next generation sequencing (NGS) technologies including genotyping by sequencing (GBS), de novo transcriptome assembly and whole genome sequencing to generate genomic resources for nonmodel tree species on the African continent. Second, we discuss landscape genomics, an emerging field in genomic science and discuss research areas in which the genomic resources of trees in Africa can be used to inform research on landscape genomics and to improve food production. Genomics resources, including reference genomes are completely lacking for African trees. Instead, most available molecular studies of trees in Africa examined a limited number of loci for addressing questions that vary along three major axes: inferring past demographic history, species delimitation/classification, and in conservation (Supporting Information Table S1). In terms of inferring evolutionary history, much of the work addressed the phylogeographic history of dispersal, diversification, vicariance disjunction, evolutionary origin and community phylogenetic structure with focus in central and southern Africa (Table S1). These studies mostly utilized chloroplast DNA regions (usually not more than six) and sometimes nuclear markers, and microsatellites for inferring past demographic history (Table S1). For species delimitation and supraspecific classification, much of the work is grounded on DNA barcoding and disentangling phylogenetic relationships (Table S1). Tree conservation has received the least attention, with existing studies focusing on delineating biodiversity hotspots (Daru et al., 2015a) and conservation genetics (Lowe et al., 2000). Although these studies provided the baseline for further studies, it is clear that the development of DNA resources for African trees is the first step toward unlocking the potential of genomics of trees on the continent. Among the common methods for collecting genomic data in nonmodel African tree species without prior molecular resources include amplified fragment length polymorphisms (AFLPs), GBS, diversity arrays technology (DArT), DArTseq, restriction site associated DNA sequencing (RADseq), and de novo transcriptome assembly (Fry et al., 2009; Elshire et al., 2011; De Wit et al., 2012; Van Schalkwyk et al., 2012; Li et al., 2015). These methods offer a moderately cheap means of collecting thousands of genomic DNA data points on multiple samples from nonmodel species without reference genomes (Hohenlohe et al., 2010; Elshire et al., 2011; Yeaman et al., 2014). The GBS technology generates data for thousands of loci per sample by reducing complexity in the genome using restriction enzymes before sequencing (Elshire et al., 2011; Beissinger et al., 2013). A precursor to the GBS was diversity arrays (DArT), a microarray platform for genome wide marker development and analysis. One of the earliest applications of DArT was in a tree species – Eucalyptus (Lezar et al., 2004), and it was subsequently widely employed for plant genotyping, often in orphan crops (Gemenet et al., 2015). The GBS version of DArT (DArTseq), provides an order of magnitude more markers (Li et al., 2015). Another variant of the GBS is RADseq, which is being implemented broadly for population genomics of nonmodel species with no prior genomic resources (Andrews et al., 2014). The advantages of these GBS methods are: (1) low cost, (2) not requiring reference genome sequence, (3) treatment of polymorphic tags as dominant markers for many research questions, and (4) can be tailored for each species (Beissinger et al., 2013). The disadvantages include: (1) bioinformatics challenges in alignment of large and polyploid genomes, (2) bias of genomic regions sampled due to the restriction enzyme–based complexity reduction, and (3) systematic errors for example PCR duplicates introduced during library preparation (Andrews et al., 2014; He et al., 2014). Another technique to develop genomic resources for African trees is to build a catalog of expressed genes using RNAseq (de novo transcriptomes; Mizrachi et al., 2010). This provides information on gene model structure as well as the temporal, spatial and developmental contexts of gene expression. Additionally, we can combine transcriptome with genetic data through expression QTL (quantitative trait locus) analysis to better understand gene regulation, for example in trees growing in different environments (e.g. Munkvold et al., 2013). Genome 'skimming' is another emerging idea that could be applied for conservation of African trees, through the extraction of high-copy sequences, such as plastid genomes, from a low coverage genome (Dodsworth, 2015). This approach may indeed replace the use of PCR-based barcoding techniques (e.g. Gere et al., 2013; Bello et al., 2015), of which the search for a universal barcode for plants is still underway (Hollingsworth et al., 2011). Finally, if financial resources were available, a good strategy would be to do whole genome sequencing of representative species per family, as carried out for Eucalyptus grandis (Myburg et al., 2014). Reference genomes facilitate the speed and accuracy of more cost-effective methods applied to genetically diverse populations, such as GBS, transcriptomics or genome resequencing at low coverage (De Wit et al., 2012). Currently, Sanger sequencing remains the gold standard for low throughput sequencing applications such as barcoding (Hollingsworth et al., 2011). However, implementation of short-read NGS technologies such as Illumina, 454 pyrosequencing or Ion Torrent in plant research is growing at an unprecedented rate (Varshney et al., 2014), and holds promise for African trees. The advantages of short-read technologies are high-throughput, cost-effectiveness, low error rate (< 1%; Reuter et al., 2015) compared to other NGS methods, and established analytical tools. For example, a commonly used genome assembler, Velvet, was developed 7 yr ago (Zerbino & Birney, 2008). A disadvantage of short-read technologies is the inability to span regions of repetitive DNA, thus hampering genome assembly (Reuter et al., 2015). Other emerging sequencing technologies that can be applied for African trees include single molecule sequencing available from Pacific Biosciences and Oxford Nanopore which can produce longer reads (usually > 10 kb) (Reuter et al., 2015). A current drawback of single molecule sequencing techniques is the error rate of 5% or more (Reuter et al., 2015). Increased sequencing coverage, for example by sequencing the same molecule multiple times (PacBio), is used to compensate for this, however this makes the cost per base pair greater than short-read NGS. Given that most of the established sequencing technologies are capital intensive, often requiring centralized service providers operating on mega scales, scientists interested in African tree genomics might be faced with the challenge of whether to prioritize technologies or hypothesis-driven research based on experimental design and data analysis. New innovations include the Oxford Nanopore MinION, which is akin to a 'lab-on-chip' (Watson et al., 2015). It is a cheap and portable (only ~10 cm long) mobile DNA sequencing device powered through a USB port of a laptop, hence ideal for fieldwork. The catalogued genomic resources derived from GBS, transcriptomics, and reference whole genome sequencing mentioned earlier can be used for addressing various questions relating to the ecology and evolution of African trees. Here we focus on the prospects of these methods in the study of landscape genomics and to improve food production in Africa. Landscape genomics is a field that derives from population genetics. It involves the concurrent study of numerous loci from a genome (usually hundreds of markers, genes or genomic regions), both neutral or under adaptive selection for each georeferenced individual across a landscape. Landscape genomics amalgamates the fields of geographic information systems and genomics to explore population dynamics and adaptive genetic variation across landscapes (Schwartz et al., 2010; Bragg et al., 2015). Landscape genomics has been explored in model plants with reference genomes such as Arabidopsis (Turner et al., 2010; Fournier-Level et al., 2011; Li et al., 2014), and recently in forest trees (Eckert et al., 2009, 2010; De Kort et al., 2014; Geraldes et al., 2014; McLean et al., 2014). In these studies a strong link between genotype, phenotype and the environment were demonstrated. Furthermore, these studies revealed that genomic data can identify gene regions under positive selection controlling adaptive phenotypic traits in response to environmental variables (see Bragg et al., 2015). The ultimate goal is to precisely quantify population dynamics such as gene flow, genetic differentiation, diversity, and their interaction with several environmental variables in the landscape (Schwartz et al., 2010; Bragg et al., 2015). Quantifying genomic regions under selective pressures by the environment for landscape genomics involves surveying many loci scattered across the genome of an individual (Manel et al., 2012). Fundamentally, this requires three types of dataset: the genomic data of the species under study, occurrence/abundance data of the species, and environmental variables (e.g. soil type, precipitation, temperature, photoperiod) in the study area (Hand et al., 2015). Climatic data are readily available, for example via WorldClim (Hijmans et al., 2005), and plant distribution data from the Global Biodiversity Information Facility (www.gbif.org). Here, we focus on synthesizing genomic data of African tree species. Since our focus is on landscape genomics of nonmodel African trees without previous genomic resources, the methods chosen should be able to separate neutral from nonneutral markers, and identify suitable environmental variables driving the selection (Foll & Gaggiotti, 2008; Holderegger & Wagner, 2008; Balkenhol et al., 2009). Previous studies have successfully utilized GBS, de novo transcriptome assembly, AFLPs, RADseq or single nucleotide polymorphisms (SNPs) to generate molecular resources for landscape genomics (Schwartz et al., 2010). As a result, the markers of each georeferenced individual studied using any of these methods can be correlated with environmental variables (e.g. substrate, precipitation, temperature, photoperiod, potential evapotranspiration), to understand the potential adaptive role of genomic regions in response to selective environmental pressure. Here we highlight, as examples, the potential application of landscape genomics for understanding vicariance biogeography, and gregariousness in certain African tree species. Previous phylogeographic studies of African trees were focused in central Africa (Table S1), but have not been evaluated in a genomic framework. Examples of African trees with potential for exploring phylogeography and landscape genomics include Vachellia karroo (= Acacia karroo) and Nymania capensis (see Fig. 1 for their phylogenetic positions). Vachellia karroo (Fabaceae) has a wide distribution (Fig. 2a), traversing several biomes, phytogeographic zones and climatic conditions, including semi-desert (Succulent- and Nama-Karoo), savanna and grassland, as well as both winter and summer-rainfall regimes (White, 1983; Mucina & Rutherford, 2006; Daru et al., 2015b; Fig. 2d). It is a typical so-called chorological and ecological transgressor, several examples which are displayed among African trees (White, 1981). With such wide geographical range, we expect that the allele frequencies in populations, for example in the winter-rainfall Succulent Karoo, would be different to that of the Highveld grassland with summer-rainfall and winter frost. This is because the Succulent- and Nama-Karoo population would be more adapted to drought tolerance, compared to the Highveld populations that would be adapted to higher rainfall and cold winters, but the genomic basis remains unknown. Study designs for applying landscape genomics to understand phylogeography of nonmodel African trees should use de novo transcriptome sequencing to build a catalog library of gene families from various tissues of individuals in the populations of interest. This could have important conservation implications. One can ask, what is the implication of spreading nonnative alleles from a specimen from a population with different environmental tolerance to a new area with another environmental condition, as is often the case with widespread indigenous African species used in horticulture? Another potential tree for exploring phylogeography using landscape genomics is N. capensis (Meliaceae) that has two disjunct distributions, one in the Northern Cape of South Africa and southern Namibia and the other in the Western Cape of South Africa (Fig. 2b). Such disjunct distributions between the south-eastern and the north-western parts of the subcontinent (a pattern displayed by several plant taxa) was thought to be a result of vicariant events associated with past climatic change (Van Wyk & Smith, 2001). These historically isolated populations may hold information on genetic adaptations resulting from long potential exposure to divergent selection pressures. Within historical biogeography, molecular information may also be useful for weighing the relative merits of hypotheses based on vicariance vs dispersal in explaining geographical distribution of genetic diversity (Avise, 2000). Moreover, molecular information may also assist in the recognition of different taxa within what is currently recognized as single species with disjunct distributions. NGS methods of de novo transcriptomes are already being used for species delimitation in temperate regions for nonmodel species, and this could be applied to African trees, with implications for discovering cryptic new infrageneric taxa. The latitudinal gradient of tree diversity has long intrigued biologists (e.g. Slik et al., 2015). The biotic bases for this are not yet fully understood, but genomics as a potential source of information needs to be explored. Many temperate tree species are wind-pollinated (anemophilous) and tend to grow in dense stands, in other words, they are gregarious. For wind pollination to be effective, individuals of a species are expected to occur in close proximity, a trait most probably achieved through the competitive suppression of growth of other species. Hence, based on casual observations of wind-pollinated trees (A. E. van Wyk, unpublished data, 2015), we consider gregariousness to be part of the syndrome for anemophily, a trait usually not considered in the literature on pollination biology (e.g. Friedman & Barrett, 2009). Thus, one would expect plant communities dominated by wind-pollinated trees to be comparatively poor in tree species diversity, as is the case in many temperate regions of the northern hemisphere (Slik et al., 2015). By contrast, trees in Africa are mostly animal-pollinated (Rodger et al., 2004), but with few exceptions of wind pollination, such as Colophospermum mopane (Fabaceae), Androstachys johnsonii (Picrodendraceae), and Spirostachys africana (Euphorbiaceae). Colophospermum mopane grows in dense, almost homogeneous stands that cover large parts of southern Africa (Fig. 2c). It is hypothesized that it exhibits gregariousness by competitively suppressing other woody plants, perhaps by, among others, releasing allelopathic compounds. This could be a good candidate for exploring landscape genomics. For instance, study designs could use AFLPs to establish the different known ecotypes of this species associated with different habitats (Makhado et al., 2014). One could also use de novo transcriptomics to build genomic resources towards understanding the genomic basis of gregariousness. This has potentially important implications in ecology, conservation and invasion plant biology (Chown et al., 2015). For example, if the goal is to understand the genomic basis of species invasion, exploring the genomics of gregariousness could lead to the discovery of genes responsible for competitive traits (including allelopathy) during colonization, spread, and suppression of other plants, as is characteristic of many invasive plants (Chown et al., 2015), such as Lantana camara and species of Pinus. Another implication in this direction is that it could lead to the establishment of a link between pollination syndrome, gregariousness and tree diversity, the existence of which is still hypothetical at this stage. The exponential population growth rate and the ongoing global climate change is impacting negatively on food production and nutrition in sub-Saharan Africa (Bloom, 2011). While African trees produce various fruits essential for maintaining animal populations (e.g. Breitwisch, 1983; Bleher et al., 2003; Kissling et al., 2008; Daru et al., 2015c), their potential value as food for the increasing human population remains untapped. As a result, genomic science can play a role in improving nutritional quality and increase food production (e.g. see Wang et al. (2015) for domestication of maize from teosinte). Thus, genomics can be applied to domestication of African trees with food value in three ways. First, plant breeders can tap into the power of low throughput DNA markers (e.g. PCR based single sequence repeats) or high throughput markers (e.g. SNPs) to select desirable genotypes or counter-select undesirable ones. Second, study designs should use transcriptome sequencing to build a catalog of expressed genes for important African trees, toward discovering genes involved in fruit quality, yield, pest and drought resistance. Third, in cases of barriers to conventional breeding (e.g. requirement for vegetative propagation), the power of genetic modification (GM) can be harnessed via genomics and gene discovery. However, care must be taken to account for on-going debates on GM crops (Gerasimova, 2015). The search is in progress for genes under positive selection for desirable traits such as increased crop yield, disease and pest resistance, drought tolerance, and sweetness (Tester & Langridge, 2010; Kanchiswamy et al., 2015). Indeed, on the African continent, one such initiative is the African Orphan Crops Consortium (AOCC; http://africanorphancrops.org/) with the aim of sequencing the genomes of >100 plants commonly used for food in Africa. Specifically, the AOCC aims to develop genomic resources for 'orphan crops' in Africa – crops which are typically neglected and underutilized in regional food security, but have important food or medicinal properties (Naylor et al., 2004). Only 31 species (out of the top 100 plant species) are African trees (Table S2), which is a promising start. However, these are sparsely distributed across the phylogenetic tree of plant life (Fig. 1), highlighting the need to expand genomics research to cover more representative members of the over 6000 tree species in Africa. Trees in Africa play keystone roles, having a disproportionately strong influence on co-occurring species, via ecosystem services, worth many millions of dollars; but they are also threatened by environmental and anthropogenic factors. Tapping the potential value of trees and how they respond to these challenges would require an integrative approach of which genomic science has a major role to play. Whereas the availability of reference genomes for many species in the temperate regions has promoted genomics research for temperate trees, here we provided ways to unlock the potential of genomics for nonmodel trees in Africa by focusing on landscape genomics and improved food production and nutrition. While our plea focuses on opportunities for exploiting genomics science for African trees because they represent one of the most defining features of the African landscapes, our suggestions can be extrapolated to other tropical regions such as South America or southeast Asia, or other taxonomic groups including epiphytes, herbs or grasses in the tropics for which genomics resources are lacking. We hope that these suggestions along with the ongoing precipitous drop in DNA sequencing costs and improvements in data quality will open new frontiers in the study of the diversity, ecology and evolution of African trees. The authors thank the University of Pretoria for logistic support. Special thanks to Courtney A. Hollender for comments on an earlier draft of the manuscript, as well as three anonymous reviewers, and to Jacques van den Berge for help with data collation. B.H.D. acknowledges financial support from the Genomics Research Institute, University of Pretoria (UP). Opinions expressed and conclusions reached, are those of the authors and not necessarily that of the UP. B.H.D. planned and designed the research. B.H.D. analyzed data. B.H.D., D.K.B. and A.E.v.W. contributed materials and analysis tools. B.H.D., D.K.B. and A.E.v.W. wrote the manuscript. Please note: Wiley Blackwell are not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing material) should be directed to the New Phytologist Central Office. Table S1 Summary of findings of some previous molecular studies of trees/woody species across Africa in the fields of phylogeography, molecular systematics and phylogenetics (e.g. DNA barcoding) Table S2 List of African trees with food value prioritized for genomics analysis by the African Orphan Crops Consortium (http://africanorphancrops.org/) Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.