Fonio millet genome unlocks African orphan crop diversity for agriculture in a changing climate

Humanity faces the challenge of sustainably producing food for a growing population (9-10 billion by 2050) amidst climate change. Efficient use of plant diversity and genetic resources in breeding is crucial to diversify and transform agriculture. Arid and semi-arid regions, particularly in Africa and the Middle East, are highly vulnerable to decreased agricultural productivity due to climate change, land degradation, and water depletion. Agricultural selection has significantly reduced agrobiodiversity, with a few cereal crops (wheat, maize, rice) dominating global calorie consumption. These major crops, originating in humid regions, have limited genetic diversity for extreme environments. In contrast, crop wild relatives and orphan crops, often adapted to harsh conditions, offer potential for unlocking marginal lands for agriculture. Advances in genomics and genome editing provide opportunities for rapid domestication of wild relatives and improvement of orphan crops. White fonio (*Digitaria exilis*) is a West African millet species with significant potential for agriculture in marginal environments. It's a fast-maturing, drought-tolerant cereal adapted to nutrient-poor soils, but possesses unfavorable traits like seed shattering and lower yields compared to other cereals. This study aims to establish comprehensive genomic resources for fonio to harness its potential for agriculture in harsh environments, including a high-quality reference genome assembly and deep re-sequencing of a diverse panel of wild and cultivated accessions.

Previous research has highlighted the importance of crop diversification for food security in the face of climate change (Hickey et al., 2019; Tena, 2019; National Academies of Sciences Engineering and Medicine, 2019; FAO, 2018). Studies have shown the vulnerability of arid and semi-arid regions to decreasing agricultural productivity (Dalin et al., 2017). The reduction in agrobiodiversity due to agricultural selection has been extensively documented (Gruber, 2017; Tanksley & McCouch, 1997; Fernie & Yan, 2019; Eshed & Lippman, 2019). The potential of orphan crops and wild relatives to improve agriculture in challenging environments has gained traction (Kantar & Runck, 2019; Dawson et al., 2019; Pironon et al., 2019). Fonio's potential as a drought-tolerant, fast-maturing cereal has been noted (Ayenan et al., 2018; Cruz & Beavogui, 2016; Adoukonou-Sagbadja et al., 2010; Abdul & Jideani, 2019), along with its undomesticated characteristics (Barnaud et al., 2017). Previous studies on fonio's reproductive system and phylogenetic relationships have provided a foundation for this work (Adoukonou-Sagbadja et al., 2007).

This study involved multiple steps: 1. **Genome size estimation:** Flow cytometry was used to estimate the genome size of fonio accession CM05836. 2. **Reference genome sequencing and assembly:** High-molecular-weight DNA was extracted from CM05836, and multiple sequencing libraries (Illumina paired-end, mate-pair, and 10x Genomics linked-read) were constructed and sequenced. De novo assembly was performed using DeNovoMAGIC3, followed by scaffolding with Hi-C data and a Bionano optical map to produce a chromosome-scale assembly. A second assembly was generated using the TRITEX pipeline for comparison. 3. **Gene annotation:** Gene annotation was performed using the MAKER pipeline, integrating RNA-Seq data from various tissues and protein sequences from other plant genomes. 4. **Synteny and comparative genome analysis:** Comparative genomic analysis was conducted with other grass species to determine syntenic relationships and divergence times. 5. **Re-sequencing of *D. exilis* and *D. longiflora* accessions:** 166 *D. exilis* and 17 *D. longiflora* accessions were selected for whole-genome re-sequencing. Sequencing reads were mapped to the reference assembly, and SNPs were called and filtered. 6. **Genetic diversity and population structure:** PCA, genetic structure analysis (sNMF), and LD decay analysis were performed to investigate genetic diversity and population structure. Associations between genetic structure and climate, geography, and ethnolinguistic factors were assessed using correlation and Mantel tests. 7. **Genome scanning for selection signals:** Three methods (CLR test, nucleotide diversity ratios, and FST calculations) were used to identify genomic regions under selection. Orthologs of known domestication genes were identified in these regions. 8. **Phenotyping:** Seed shattering was measured in *D. exilis* accessions with and without a deletion in the *DeSh1-9A* gene.

A high-quality, chromosome-scale reference genome assembly of fonio (716.47 Mb) was generated. Analysis of full-length LTR retrotransposons indicated that the fonio genome is a recent allotetraploid (∼3 MYA), with two subgenomes (A and B) showing similar numbers of orthologous genes and expression levels. Analysis of 183 *Digitaria* accessions revealed significant genetic diversity shaped by climatic, geographic, and ethnolinguistic factors. Genome-wide association studies identified loci associated with adaptation to temperature and precipitation, and with ethnic groups. A genetic bottleneck was detected in *D. exilis*, likely linked to domestication. Selection scans revealed two genes under selection: *DeGS5-3A* (grain size) showed a complete loss of diversity in the coding sequence in *D. exilis*, and *DeSh1-9A* (seed shattering) showed a partial sweep, with a 60 kb deletion present in 37% of *D. exilis* accessions. The deletion resulted in a small but significant reduction in seed shattering. Most known domestication genes from other cereals showed no strong selection in fonio.

This study provides valuable genomic resources for fonio improvement. The chromosome-scale assembly and deep re-sequencing data provide a foundation for molecular breeding efforts. The identification of *DeGS5-3A* and *DeSh1-9A* as targets for improving grain size and shattering represents significant progress. The lack of strong selection on most other domestication genes from other cereals suggests potential for rapid improvement by focusing on these key genes. The impact of geographic, climatic, and human factors on shaping fonio diversity highlights the importance of considering these factors in breeding programs. The mild bottleneck observed in fonio compared to other crops may facilitate breeding efforts.

This study presents a comprehensive set of genomic resources for fonio millet, facilitating future molecular breeding efforts. The identification of key genes under selection, along with insights into population structure and domestication history, provides valuable targets for improving this underutilized crop. Future research should focus on leveraging these resources for developing improved fonio cultivars adapted to various environments and farmer preferences. Further investigation into the genetic basis of other key traits, such as yield and disease resistance, will be crucial.

The study focused on a specific set of *D. exilis* and *D. longiflora* accessions, potentially limiting the generalizability of some findings. The association analyses relied on existing ethnolinguistic data, which might not capture the full complexity of social factors influencing fonio diversity. Further phenotyping data on a broader range of accessions would strengthen the selection scan results. Future studies should consider additional environmental variables and a more extensive range of accessions to enhance the results.