On the accuracy of threshold genomic prediction models for leaf miner and leaf rust resistance in arabica coffee.

• Autor(a) principal: CARVALHO, H. F.
• Data de Publicação: 2023
• Outros Autores: FERRÃO, L. F. V., GALLI, G., NONATO, J. V. A., PADILHA, L., MALUF, M. P., RESENDE JR., M. F. R. de, FRITSCHE-NETO, R., GUERREIRO-FILHO, O.
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Institucional da EMBRAPA (Repository Open Access to Scientific Information from EMBRAPA - Alice)
• Texto Completo: http://www.alice.cnptia.embrapa.br/alice/handle/doc/1156023; https://doi.org/10.1007/s11295-022-01581-8
• topic: Coffea arabica var. arabica; Leucoptera; Hemileia; Leaf rust; Genomics

Obtaining resistance cultivars for leaf miner and leaf rust are the main important strategy of Brazil?s national coffee breeding program. The narrow genetic basis, and founder effect consequences, lead to challenges in quantifying and detecting genetic diversity for these traits. Biotechnology tools allied with classical breeding strategies are powerful in detecting variability and deploying a precision selection. The selection based on the genetic merit of an individual obtained from thousands of single nucleotide polymorphism effects is known as genomic selection. The ordinal scale principally makes the resistance evaluation of the leaf rust and leaf miner of the score, categorizing the phenotypes following the discrete (ordinal) distribution. Hence, this distribution can be better analyzed by threshold models. Our goals were to optimize genomic prediction models for coffee resistance to leaf rust and leaf miner via threshold models and compare pedigree and genomic relationship matrices to underlying prediction models. We have observed that the genomic model with the genomic relationship matrix performed better for all scenarios. For the traits with at least five degrees of scores, the threshold models performed better, whereas for a trait with ten degrees of scores, we see no advantage to using a threshold model for genomic prediction.