Performing Department
(N/A)
Non Technical Summary
The honey bee,Apis mellifera, is the principal pollinator for $215 billion worth of crops worldwide each year. Despite high demand for their pollination service, record losses have severely limited beekeepers' ability to supply sufficient colonies. A sustainable, long-term solution requires genetic improvements in growth; pest, pathogen, and disease resistance; and honey yield. Commercial honey bee breeders have struggled to produce significant gains in all these areas over the past century largely due to limitations in time, but also economic cost and the logistic constraints of integrating multi-trait selection into commercial honey bee breeding operations. Genomic selection - a proven technology in animal and plant breeding - significantly reduces the effect of these limitations while providing increased genetic gain. If adapted,genomic selectionhas profound potential to improve honey bee breeding.What if commercial honey bee breeders could survey their mating populations and identify individuals that would produce desirable outcomes prior to having to establish and maintain colonies? Genomic selection is at the tipping point where a seamless integration into existing commercial applications can be readily implemented offering significant short-term genetic information leading to sustainable long-term breeding solutions to major honey bee threats.I willdevelop a high-throughput, rapid, and inexpensive method for stakeholders to incorporate genomic selection into their breeding programs.
Animal Health Component
30%
Research Effort Categories
Basic
20%
Applied
30%
Developmental
50%
Goals / Objectives
The limit imposed by phenotyping is not unique to honey bees; similar issues have been described in other animal breeding programs where genomic selection has emerged as a critical tool.Genomic selection provides significant accuracy and predictive power across multiple traits which has accelerated genetic gains in various livestock. Specifically, a genomic selection approach establishes breeding values in a reference population by modeling correlations of phenotypes with molecular markers across the genome. Once developed, the model can be used to estimate the breeding potential of un-phenotyped individuals using affordable, accessible sequencing in lieu of costly phenotyping. This approach is particularly valuable for complex or multi-genic traits such as growth, pest and disease resistance, and yield. The proposed constitutes the first effort to deploy genomic selection in honey bee breeding towards enabling honey bee breeders with greater ability to evaluate colonies and traits in a cost-effective manner.This project aims to create a genomic selection methodology specific to commercial honey bee breeders, evaluating ability to predict traits and enhance genetic gains in a large scale, high throughput manner.Objective 1: Characterize population genetics of the reference population.Rationale:Genomic selection relies on the ability to develop a reference panel from a population.Accuracy of predictions from genomic selection is alsoinfluenced by genetic diversity and population structure. Greater genetic diversity within the reference population provides increased certainty in the discovery of candidate loci for breeding values, it also allows genomic selection potential to persist over several generations, and allows predictive ability to all 17 honey bee breeders. Unaccounted population structure can make it difficult to predict phenotypes across multiple breeders. It is critical to characterize genetic diversity and population structure of the referenced population to establish a robust genomic selection approach.I expect the reference population will have high genetic diversity but will not be structured due tomigratory nature of beekeeping.Objective 2: Evaluate prediction accuracy of genomic selection models.Genomic breeding values are essential for honey bee breeders, as they help predict which queens are likely to produce offspring with desirable traits. This accuracy enables breeders to select queens based on these values, significantly reducing the time required to achieve significant genetic improvements in their breeding population. Genomic selection is particularly valuable for multi-trait selection, especially when assessing traits that are challenging and costly to evaluate in a large number of honey bees - a current limitation faced by the 17 honey bee breeders. In practice, breeders must consider multiple traits simultaneously when selecting improved genetic stock. Consequently, understanding the genetic correlations among 20 phenotypes is crucial and may be integrated into a multi-trait genomic selection framework. The second objective involves developing and comparing multiple genomic models from the reference population of 17 honey bee breeders to determine their impact on genomic selection prediction accuracy.I predict that multi-trait genomic selection models will yield the highest prediction accuracy.Objective 3: Identify breeding schemes for genomic selection.The decreasing cost of genotyping and the growing interest from honey bee breeders have made genomic selection more feasible. As a result, it is crucial to establish guiding principles for the 17 honey bee breeders to incorporate this technology into their operations. Objective 3 will concentrate on pinpointing strategies that breeders can adopt to integrate genomic selection. To optimize breeding schemes, we will use simulation models based on real and simulated data (Objectives 1 and 2). This data will allow us to compare existing breeding schemes with genomic selection schemes, assisting us in determining the best strategies for each operation. We predict that specific strategies (i.e., higher screening, employing mating strategies) will improve both traditional and genomic breeding methods, with genomic selection outperforming conventional methods for every breeder.?
Project Methods
Objective 1: Characterize population genetics of the reference population.I will measure 20 heritable industry-relevant traits on 1,000 colonies across North Dakota over a two-year period to establish and characterize a comprehensive reference population. Thirty colonies from each beekeeper will be phenotyped twice (Mid-May and Mid-August 2024) during a period when colonies are unmanaged due to the North Dakota honey flow. The proposed three month period allows for accurate estimation of colony growth, health, survival, honey yield, and pest, pathogen, disease load without the advent of management. Measures of colony growth and pest/pathogen/disease prevalence during this period in North Dakota (Mid-May and Mid-August) are known to be predictive of colony health, survival, and economic value of colonies during pollination services. Then, population structure will be investigated in the reference population using three complementary approaches: 1) an admixture analysis provides insight into the genetic ancestry(42), 2) a Principal Component Analysis (PCA) and Identity by Descent (IBD) estimates will measure genetic stratifications(27), and 3) genome-wide Fst estimates genetic differentiation(27).Objective 2: Estimate prediction accuracy of genomic selection modelsI will compare both multi-trait and single-trait genomic selection models to assess prediction accuracy genomic breeding values using the genotype and phenotype data from Objective 1. Objective 2 involves comparing various statistical genome-based models, such as Genomic Best Linear Unbiased Prediction (GBLUP). These models will be fitted with queen, male, worker, or colony genetic effects and will utilize genetic groups to account for unknown honey bee males (GBLUP). Negative values will be assigned to traits where a decrease represents improvement (i.e., pest/pathogens/diseases). We plan to evaluate four single-trait Genomic Selection models: GBLUP, Bayesian ridge regression (BRR), BayesCπ, and the threshold GBLUP designed for ordinal traits. Additionally, we will assess Genomic Selection models that incorporate information on multiple correlated traits into a single analysis using GBLUP multivariate models. To evaluate the prediction accuracy of genomic selection models, datasets (phenotypic and genomic) will undergo cross-validation by splitting them into two groups (training and validation populations), with 95%, 50%, or 10% of colonies in the training population and the remaining 5%, 50%, or 90% in the validation population. Multiple analyses will be conducted for each combination of trait, genomic selection model, replicates (20), and training population size (3 sizes). Prediction accuracy will be calculated to determine which model most accurately predicts genomic breeding values. For multi-trait models, the prediction ability (PA) and prediction accuracy (PACC) of multi-trait GBLUP models will be compared with their single-trait GBLUP counterparts, using adjusted phenotypes as response variables.Objective 3: Identify breeding schemes for genomic selection.The following strategies will be tested. Conventional and Genomic breeding schema will be produced based on each honey bee breeders' current practice, including selection intensity, generation interval, accuracy of breeding values, and genetic diversity within their operation. We will then compare genetic progress between their current strategy and alternative strategies, such as selection at an earlier stage, II, and phenotyping and genotyping a larger sample size. Alternative Genomic Selection models (Objective 2) will be tested for predicting genomic breeding values. Alternative genotyping approaches will be investigated by comparing low- coverage to high-coverage sequencing (0.01x to 30x). Low-coverage sequencing can be a cost-effective option for large-scale genomic selection applications. The benefits of adding closed-mating will be investigated. In practical scenarios, this can be achieved using II and geographical isolation of breeding colonies. The advantages of each strategy will be evaluated based on the increase in genetic progress per generation versus the economic and practical feasibility of implementing such practices based on each honey bee breeders operation.