Quantitative assessment of observed versus predicted responses to selection

Pélabon, Christophe; Albertsen, Elena; Rouzic, Arnaud Le; Firmat, Cyril; Bolstad, Geir H.; Armbruster, W. Scott; Hansen, Thomas F.

doi:10.1111/evo.14284

Cited by 11 publications

(15 citation statements)

References 96 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The problem with this approach is that both the breeder’s prediction and the change in the mean for the trait are measured with noise, which typically is very large. This random component of the prediction error is due to several factors, including drift and sampling, and can be so large as to obscure the bias estimated using only the past generation ( 30 ). Furthermore, the stochastic component of the error is independent in each generation, so it contains no useful information that can be exploited to improve predictions.…”

Section: Resultsmentioning

confidence: 99%

A method to predict the response to directional selection using a Kalman filter

Milocco

Salazar‐Ciudad

2022

Proc. Natl. Acad. Sci. U.S.A.

View full text Add to dashboard Cite

Predicting evolution remains challenging. The field of quantitative genetics provides predictions for the response to directional selection through the breeder’s equation, but these predictions can have errors. The sources of these errors include omission of traits under selection, inaccurate estimates of genetic variance, and nonlinearities in the relationship between genetic and phenotypic variation. Previous research showed that the expected value of these prediction errors is often not zero, so predictions are systematically biased. Here, we propose that this bias, rather than being a nuisance, can be used to improve the predictions. We use this to develop a method to predict evolution, which is built on three key innovations. First, the method predicts change as the breeder’s equation plus a bias term. Second, the method combines information from the breeder’s equation and from the record of past changes in the mean to predict change using a Kalman filter. Third, the parameters of the filter are fitted in each generation using a learning algorithm on the record of past changes. We compare the method to the breeder’s equation in two artificial selection experiments, one using the wing of the fruit fly and another using simulations that include a complex mapping of genotypes to phenotypes. The proposed method outperforms the breeder’s equation, particularly when traits under selection are omitted from the analysis, when data are noisy, and when additive genetic variance is estimated inaccurately or not estimated at all. The proposed method is easy to apply, requiring only the trait means over past generations.

show abstract

Section: Resultsmentioning

confidence: 99%

A method to predict the response to directional selection using a Kalman filter

Milocco

Salazar‐Ciudad

2022

Proc. Natl. Acad. Sci. U.S.A.

View full text Add to dashboard Cite

show abstract

“…This review emphasizes the potential complexity of fine-grained interactions among genotypes of different crop species. Coupled with the generally low predictability of short-term selection response in plants ( Pélabon et al, 2021 ), this should encourage the search for robust qualitative evidence to facilitate the choice of breeding strategies. The overview we provided (summarized in Table 1 ) of the available models and their parameters should be useful in this regard.…”

Section: Discussionmentioning

confidence: 99%

“…In cases where the magnitude of the genetic covariance between DGE and IIGE is non-null, the evolution of this first trait is expected to affect the magnitude of IIGE on other species' traits, hence the mean value of other species traits. The notation σ DGE A1 ,IIGGE A1|B1 , for instance, denotes the covariance between the DGE of species A associated with trait A 1 vs. its IIGE affecting the trait B 1 of species B. correlated response to selection (Gromko, 1995), and obtaining reliable predictions of the correlated response to recurrent artificial selection in plants seems very challenging (Pélabon et al, 2021). It has also been shown that estimates of speciesspecific G-matrices are environment-dependent (Wood and Brodie, 2015), i.e., the environment in which they are measured influences their estimation, which can introduce a strong bias in predictions that were made for another environment.…”

Section: Cross-species Genetic Correlations and Inaccurate Quantitati...mentioning

confidence: 99%

“…It will be recalled that when restricted to traits measured in the same species, genetic correlations (a standard measure for genetic covariances) are often the poor predictors of the actual correlated response to selection ( Gromko, 1995 ), and obtaining reliable predictions of the correlated response to recurrent artificial selection in plants seems very challenging ( Pélabon et al, 2021 ). It has also been shown that estimates of species-specific G-matrices are environment-dependent ( Wood and Brodie, 2015 ), i.e., the environment in which they are measured influences their estimation, which can introduce a strong bias in predictions that were made for another environment.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Linking quantitative genetics with community-level performance: Are there operational models for plant breeding?

Firmat

Litrico²

2022

Front. Plant Sci.

Self Cite

View full text Add to dashboard Cite

Plant breeding is focused on the genotype and population levels while targeting effects at higher levels of biodiversity, from crop covers to agroecosystems. Making predictions across nested levels of biodiversity is therefore a major challenge for the development of intercropping practices. New prediction tools and concepts are required to design breeding strategies with desirable outcomes at the crop community level. We reviewed theoretical advances in the field of evolutionary ecology to identify potentially operational ways of predicting the effects of artificial selection on community-level performances. We identified three main types of approaches differing in the way they model interspecific indirect genetic effects (IIGEs) at the community level: (1) The community heritability approach estimates the variance for IIGE induced by a focal species at the community level; (2) the joint phenotype approach quantifies genetic constraints between direct genetic effects and IIGE for a set of interacting species; (3) the community-trait genetic gradient approach decomposes the IIGE for a focal species across a multivariate set of its functional traits. We discuss the potential operational capacities of these approaches and stress that each is a special case of a general multitrait and multispecies selection index. Choosing one therefore involves assumptions and goals regarding the breeding target and strategy. Obtaining reliable quantitative, community-level predictions at the genetic level is constrained by the size and complexity of the experimental designs usually required. Breeding strategies should instead be compared using theoretically informed qualitative predictions. The need to estimate genetic covariances between traits measured both within and among species (for IIGE) is another obstacle, as the two are not determined by the exact same biological processes. We suggest future research directions and strategies to overcome these limits. Our synthesis offers an integrative theoretical framework for breeders interested in the genetic improvement of crop communities but also for scientists interested in the genetic bases of plant community functioning.

show abstract

“…Indeed, when applied to real systems, violations of the assumptions of the breeder's equation lead to prediction errors (Gimelfarb and Willis 1994, Rice 2004, Roff 2007, Morrissey et al 2010, Pujol et al 2018, Walsh and Lynch 2018, Shaw 2019, Milocco and Salazar-Ciudad 2020, Pélabon et al 2021. A notable example is the problem of stasis (Merilä et al 2001, Shaw 2019 where no response to selection is observed in a population that both has ample additive genetic variance and is under strong directional selection.…”

Section: Introductionmentioning

confidence: 99%

Learning from your mistakes: a novel method to predict the response to directional selection

Milocco¹,

Salazar‐Ciudad²

2021

Preprint

View full text Add to dashboard Cite

Predicting how populations respond to selection is a key goal of evolutionary biology. The field of quantitative genetics provides predictions for the response to directional selection through the breeder’s equation. However, differences between the observed responses to selection and those predicted by the breeder’s equation occur. The sources of these errors include omission of traits under selection, inaccurate estimates of genetic variance, and nonlinearities in the relationship between genetic and phenotypic variation. A key insight from previous research is that the expected value of these prediction errors is often not zero, in which case the predictions are systematically biased. Here, we propose that this prediction bias, rather than being a nuisance, can be used to improve the predictions. We use this to develop a novel method to predict the response to selection, which is built on three key innovations. First, the method predicts change as the breeder’s equation plus a bias term. Second, the method combines information from the breeder’s equation and from the record of past changes in the mean, to estimate the bias and predict change using a Kalman filter. Third, the parameters of the filter are fitted in each generation using a machine-learning algorithm on the record of past changes. We apply the method to data of an artificial selection experiment of the wing of the fruit fly, as well as to an in silico evolution experiment for teeth. We find that the method outperforms the breeder’s equation, and notably provides good predictions even when traits under selection are omitted from the analysis and when additive genetic variance is estimated inaccurately. The proposed method is easy to apply since it only requires recording the mean of the traits over past generations.

show abstract

Quantitative assessment of observed versus predicted responses to selection

Cited by 11 publications

References 96 publications

A method to predict the response to directional selection using a Kalman filter

A method to predict the response to directional selection using a Kalman filter

Linking quantitative genetics with community-level performance: Are there operational models for plant breeding?

Learning from your mistakes: a novel method to predict the response to directional selection

Contact Info

Product

Resources

About