Advances in genomic characterization of Urochloa humidicola: exploring polyploid inheritance and apomixis.

KEY MESSAGE: We present the highest-density genetic map for the hexaploid Urochloa humidicola. SNP markers expose genetic organization, reproduction, and species origin, aiding polyploid and tropical forage research. Tropical forage grasses are an important food source for animal feeding, with Urochloa humidicola, also known as Koronivia grass, being one of the main pasture grasses for poorly drained soils in the tropics. However, genetic and genomic resources for this species are lacking due to its genomic complexity, including high heterozygosity, evidence of segmental allopolyploidy, and reproduction by apomixis. These complexities hinder the application of marker-assisted selection (MAS) in breeding programs. Here, we developed the highest-density linkage map currently available for the hexaploid tropical forage grass U. humidicola. This map was constructed using a biparental F1 population generated from a cross between the female parent H031 (CIAT 26146), the only known sexual genotype for the species, and the apomictic male parent H016 (BRS cv. Tupi). The linkage analysis included 4873 single nucleotide polymorphism (SNP) markers with allele dosage information. It allowed mapping of the ASGR locus and apospory phenotype to linkage group 3, in a region syntenic with chromosome 3 of Urochloa ruziziensis and chromosome 1 of Setaria italica. We also identified hexaploid haplotypes for all individuals, assessed the meiotic configuration, and estimated the level of preferential pairing in parents during the meiotic process, which revealed the autopolyploid origin of sexual H031 in contrast to apomictic H016, which presented allopolyploid behavior in preferential pairing analysis. These results provide new information regarding the genetic organization, mode of reproduction, and allopolyploid origin of U. humidicola, potential SNPs markers associated with apomixis for MAS and resources for research on polyploids and tropical forage grasses.

A joint learning approach for genomic prediction in polyploid grasses.

Poaceae, among the most abundant plant families, includes many economically important polyploid species, such as forage grasses and sugarcane (Saccharum spp.). These species have elevated genomic complexities and limited genetic resources, hindering the application of marker-assisted selection strategies. Currently, the most promising approach for increasing genetic gains in plant breeding is genomic selection. However, due to the polyploidy nature of these polyploid species, more accurate models for incorporating genomic selection into breeding schemes are needed. This study aims to develop a machine learning method by using a joint learning approach to predict complex traits from genotypic data. Biparental populations of sugarcane and two species of forage grasses (Urochloa decumbens, Megathyrsus maximus) were genotyped, and several quantitative traits were measured. High-quality markers were used to predict several traits in different cross-validation scenarios. By combining classification and regression strategies, we developed a predictive system with promising results. Compared with traditional genomic prediction methods, the proposed strategy achieved accuracy improvements exceeding 50%. Our results suggest that the developed methodology could be implemented in breeding programs, helping reduce breeding cycles and increase genetic gains.

A Semi-Automated SNP-Based Approach for Contaminant Identification in Biparental Polyploid Populations of Tropical Forage Grasses.

Artificial hybridization plays a fundamental role in plant breeding programs since it generates new genotypic combinations that can result in desirable phenotypes. Depending on the species and mode of reproduction, controlled crosses may be challenging, and contaminating individuals can be introduced accidentally. In this context, the identification of such contaminants is important to avoid compromising further selection cycles, as well as genetic and genomic studies. The main objective of this work was to propose an automated multivariate methodology for the detection and classification of putative contaminants, including apomictic clones (ACs), self-fertilized individuals, half-siblings (HSs), and full contaminants (FCs), in biparental polyploid progenies of tropical forage grasses. We established a pipeline to identify contaminants in genotyping-by-sequencing (GBS) data encoded as allele dosages of single nucleotide polymorphism (SNP) markers by integrating principal component analysis (PCA), genotypic analysis (GA) measures based on Mendelian segregation, and clustering analysis (CA). The combination of these methods allowed for the correct identification of all contaminants in all simulated progenies and the detection of putative contaminants in three real progenies of tropical forage grasses, providing an easy and promising methodology for the identification of contaminants in biparental progenies of tetraploid and hexaploid species. The proposed pipeline was made available through the polyCID Shiny app and can be easily coupled with traditional genetic approaches, such as linkage map construction, thereby increasing the efficiency of breeding programs.

Convolutional Neural Networks to Estimate Dry Matter Yield in a Guineagrass Breeding Program Using UAV Remote Sensing.

Forage dry matter is the main source of nutrients in the diet of ruminant animals. Thus, this trait is evaluated in most forage breeding programs with the objective of increasing the yield. Novel solutions combining unmanned aerial vehicles (UAVs) and computer vision are crucial to increase the efficiency of forage breeding programs, to support high-throughput phenotyping (HTP), aiming to estimate parameters correlated to important traits. The main goal of this study was to propose a convolutional neural network (CNN) approach using UAV-RGB imagery to estimate dry matter yield traits in a guineagrass breeding program. For this, an experiment composed of 330 plots of full-sib families and checks conducted at Embrapa Beef Cattle, Brazil, was used. The image dataset was composed of images obtained with an RGB sensor embedded in a Phantom 4 PRO. The traits leaf dry matter yield (LDMY) and total dry matter yield (TDMY) were obtained by conventional agronomic methodology and considered as the ground-truth data. Different CNN architectures were analyzed, such as AlexNet, ResNeXt50, DarkNet53, and two networks proposed recently for related tasks named MaCNN and LF-CNN. Pretrained AlexNet and ResNeXt50 architectures were also studied. Ten-fold cross-validation was used for training and testing the model. Estimates of DMY traits by each CNN architecture were considered as new HTP traits to compare with real traits. Pearson correlation coefficient r between real and HTP traits ranged from 0.62 to 0.79 for LDMY and from 0.60 to 0.76 for TDMY; root square mean error (RSME) ranged from 286.24 to 366.93 kg·ha-1 for LDMY and from 413.07 to 506.56 kg·ha-1 for TDMY. All the CNNs generated heritable HTP traits, except LF-CNN for LDMY and AlexNet for TDMY. Genetic correlations between real and HTP traits were high but varied according to the CNN architecture. HTP trait from ResNeXt50 pretrained achieved the best results for indirect selection regardless of the dry matter trait. This demonstrates that CNNs with remote sensing data are highly promising for HTP for dry matter yield traits in forage breeding programs.

Deep Learning Applied to Phenotyping of Biomass in Forages with UAV-Based RGB Imagery.

Monitoring biomass of forages in experimental plots and livestock farms is a time-consuming, expensive, and biased task. Thus, non-destructive, accurate, precise, and quick phenotyping strategies for biomass yield are needed. To promote high-throughput phenotyping in forages, we propose and evaluate the use of deep learning-based methods and UAV (Unmanned Aerial Vehicle)-based RGB images to estimate the value of biomass yield by different genotypes of the forage grass species Panicum maximum Jacq. Experiments were conducted in the Brazilian Cerrado with 110 genotypes with three replications, totaling 330 plots. Two regression models based on Convolutional Neural Networks (CNNs) named AlexNet and ResNet18 were evaluated, and compared to VGGNet-adopted in previous work in the same thematic for other grass species. The predictions returned by the models reached a correlation of 0.88 and a mean absolute error of 12.98% using AlexNet considering pre-training and data augmentation. This proposal may contribute to forage biomass estimation in breeding populations and livestock areas, as well as to reduce the labor in the field.

GISH-based comparative genomic analysis in Urochloa P. Beauv.

The genus Urochloa P. Beauv. [syn. Brachiaria (Trin.) Griseb.] comprises species of great economic relevance as forages. The genomic constitution for the allotetraploid species Urochloa brizantha (cv. Marandu) and Urochloa decumbens (cv. Basilisk) and the diploid Urochloa ruziziensis was previously proposed as BBB1B1, B1B1B2B2 and B2B2, respectively. Evidence indicates U. ruziziensis as the ancestral donor of genome B2 in U. decumbens allotetraploidy, but the origin of the genomes B and B1 is still unknown. There are diploid genotypes of U. brizantha and U. decumbens that may be potential ancestors of the tetraploids. The aim of this study was to determine the genomic constitution and relationships between genotypes of U. brizantha (2x and 4x), U. decumbens (2x and 4x) and U. ruziziensis (2x) via genomic in situ hybridization (GISH). Additionally, chromosome number and genome size were verified for the diploid genotypes. The diploids U. brizantha and U. decumbens presented 2n = 2x = 18 chromosomes and DNA content of 1.79 and 1.44 pg, respectively. The GISH analysis revealed high homology between the diploids U. brizantha and U. decumbens, which suggests relatively short divergence time. The GISH using genomic probes from the diploid accessions on the tetraploid accessions' chromosomes presented similar patterns, highlighting the genome B1 present in both of the tetraploids. Based on GISH results, the genomic constitution was proposed for the diploid genotypes of U. brizantha (B1B1) and U. decumbens (B1'B1') and both were pointed as donors of genome B1 (or B1'), present in the allotetraploid genotypes.

Corrigendum: Genetic Mapping With Allele Dosage Information in Tetraploid Urochloa decumbens (Stapf) R. D. Webster Reveals Insights Into Spittlebug (Notozulia entreriana Berg) Resistance.

[This corrects the article DOI: 10.3389/fpls.2019.00092.].

Genetic Mapping With Allele Dosage Information in Tetraploid Urochloa decumbens (Stapf) R. D. Webster Reveals Insights Into Spittlebug (Notozulia entreriana Berg) Resistance.

Urochloa decumbens (Stapf) R. D. Webster is one of the most important African forage grasses in Brazilian beef production. Currently available genetic-genomic resources for this species are restricted mainly due to polyploidy and apomixis. Therefore, crucial genomic-molecular studies such as the construction of genetic maps and the mapping of quantitative trait loci (QTLs) are very challenging and consequently affect the advancement of molecular breeding. The objectives of this work were to (i) construct an integrated U. decumbens genetic map for a full-sibling progeny using GBS-based markers with allele dosage information, (ii) detect QTLs for spittlebug (Notozulia entreriana) resistance, and (iii) seek putative candidate genes involved in defense against biotic stresses. We used the Setaria viridis genome a reference to align GBS reads and selected 4,240 high-quality SNP markers with allele dosage information. Of these markers, 1,000 were distributed throughout nine homologous groups with a cumulative map length of 1,335.09 cM and an average marker density of 1.33 cM. We detected QTLs for resistance to spittlebug, an important pasture insect pest, that explained between 4.66 and 6.24% of the phenotypic variation. These QTLs are in regions containing putative candidate genes related to defense against biotic stresses. Because this is the first genetic map with SNP autotetraploid dosage data and QTL detection in U. decumbens, it will be useful for future evolutionary studies, genome assembly, and other QTL analyses in Urochloa spp. Moreover, the results might facilitate the isolation of spittlebug-related candidate genes and help clarify the mechanism of spittlebug resistance. These approaches will improve selection efficiency and accuracy in U. decumbens molecular breeding and shorten the breeding cycle.

Expected Genotype Quality and Diploidized Marker Data from Genotyping-by-Sequencing of Urochloa spp. Tetraploids.

CORE IDEAS: Introduced concept of expected genotype quality (EGQ) and software to calculate it Provided read depth guidelines for GBS in tetraploids Developed software to generate diploidized genotype calls from VCF files Demonstrated value of aligning GBS reads to a mock reference genome for SNP discovery Recommend filtering based on GQ and read depth to prevent genotype bias Although genotyping-by-sequencing (GBS) is a well-established marker technology in diploids, the development of best practices for tetraploid species is a topic of current research. We determined the theoretical relationship between read depth and the phred-scaled probability of genotype misclassification conditioned on the true genotype, which we call expected genotype quality (EGQ). If the GBS method has 0.5% allelic error, then 17 reads are needed to classify simplex tetraploids as heterozygous with 95% accuracy (EGQ = 13) vs. 61 reads to determine allele dosage. We developed an R script to convert tetraploid GBS data in variant call format (VCF) into diploidized genotype calls and applied it to 267 interspecific hybrids of the tetraploid forage grass Urochloa. When reads were aligned to a mock reference genome created from GBS data of the Urochloa brizantha (Hochst. ex A. Rich.) R. D. Webster cultivar Marandu, 25,678 biallelic single nucleotide polymorphism (SNPs) were discovered, compared with ∼3000 SNPs when aligning to the closest true reference genomes, Setaria viridis (L.) P. Beauv. and S. italica (L.) P. Beauv. Cross-validation revealed that missing genotypes were imputed by the random forest method with a median accuracy of 0.85 regardless of heterozygote frequency. Using the Urochloa spp. hybrids, we illustrated how filtering samples based only on genotype quality (GQ) creates genotype bias; a depth threshold based on EGQ is also needed regardless of whether genotypes are called using a diploidized or allele dosage model.