A stage‐wise approach for the analysis of multi‐environment trials
The DOI matches its Crossref record; 30 of the 30 references it cites were found in a registry or an index.
Abstract
Plant breeders and variety testing agencies routinely test candidate genotypes (crop varieties, lines, test hybrids) in multiple environments. Such multi‐environment trials can be efficiently analysed by mixed models. A single‐stage analysis models the entire observed data at the level of individual plots. This kind of analysis is usually considered as the gold standard. In practice, however, it is more convenient to use a two‐stage approach, in which experiments are first analysed per environment, yielding adjusted means per genotype, which are then summarised across environments in the second stage. Stage‐wise approaches suggested so far are approximate in that they cannot fully reproduce a single‐stage analysis, except in very simple cases, because the variance–covariance matrix of adjusted means from individual environments needs to be approximated by a diagonal matrix. This paper proposes a fully efficient stage‐wise method, which carries forward the full variance–covariance matrix of adjusted means from the individual environments to the analysis across the series of trials. Provided the variance components are known, this method can fully reproduce the results of a single‐stage analysis. Computations are made efficient by a diagonalisation of the residual variance–covariance matrix, which necessitates a corresponding linear transformation of both the first‐stage estimates (e.g. adjusted means and regression slopes for plot covariates) and the corresponding design matrices for fixed and random effects. We also exemplify the extension of the general approach to a three‐stage analysis. The method is illustrated using two datasets, one real and the other simulated. The proposed approach has close connections with meta‐analysis, where environments correspond to centres and genotypes to medical treatments. We therefore compare our theoretical results with recently published results from a meta‐analysis.
Overview
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
In the citing paper
“This approach can lead to a loss of information and less effective extraction of insights from the complex MET datasets.”
From Multi-environment trials data analysis: linear mixed model-based approaches using spatial and factor analytic models (2025) · cites this paper as Piepho et al., 2012
In the citing paper
“Furthermore, the ANOVA-based analysis commonly follows a two-stage modeling approach. This involves first analyzing the data within each environment, followed by a second stage of combining the results across environments for the G × E interaction analysis (Piepho et al., 2012; Smith et al., 2005).”
From Multi-environment trials data analysis: linear mixed model-based approaches using spatial and factor analytic models (2025) · cites this paper as Piepho et al., 2012
Citations
Cites
TASSEL: software for association mapping of complex traits in diverse samples
Bradbury, P. J., Zhang, Z., Kroon, D. E., Casstevens, T. M., Ramdoss, Y., & Buckler, E. S. (2007) · Bioinformatics · doi:10.1093/bioinformatics/btm308 · Cited by 7,759 · cited 1×
“…that the focus of the analysis will be on fixed genotypic effects βa, reflecting population structure and effects for the currently scanned marker, and random genotypic effects ua that have correlation matrix depending on kinship (QK approach of Yu et al., 2006; also see Bradbury et al., 2007).”
BackgroundYu, J., Pressoir, G., Briggs, W. H., Vroh Bi, I., Yamasaki, M., Doebley, J. F., McMullen, M. D., Gaut, B. S., Nielsen, D. M., Holland, J. B., et al. (2006) · Nature Genetics · doi:10.1038/ng1702 · Cited by 4,133 · cited 5×
“For breeding value estimation (Bernardo, 1994), genomic selection (Bernardo and Yu, 2007; Piepho, 2009) or association mapping (Yu et al., 2006; Müller et al., 2011), it is convenient to first summarise the phenotypic data across environments by computing adjusted means for each genotype, and then…”
BackgroundMethodInfluentialSearle, S. R., Casella, G., & McCulloch, C. E. (1992) · Wiley Series in Probability and Statistics · doi:10.1002/9780470316856 · Cited by 1,926 · cited 3×
“2.2 The general model for two-stage analysis In this section and in Sections 2.3 and 2.4, we assume, without loss of generality, that genotype is a fixed factor, so the genotype main effect is fixed, while its interaction with the environment is random (Searle et al., 1992).”
BackgroundMethodInfluentialGGE Biplot Analysis: A Graphical Tool for Breeders, Geneticists, and Agronomists
Yan, W., & Kang, M. S. (2019) · Medical Entomology and Zoology · Cited by 1,185
Prospects for Genomewide Selection for Quantitative Traits in Maize
Bernardo, R., & Yu, J. (2007) · Crop Science · doi:10.2135/cropsci2006.11.0690 · Cited by 868 · cited 2×
“In some applications such as genomic selection and association mapping (Yu et al., 2006; Bernardo and Yu, 2007), MET are used in cross-validation, i.e. the same mixed model needs to be fitted many times, which amplifies the computational burden of single-stage analysis, making two-stage analysis…”
BackgroundMethodBLUP for phenotypic selection in plant breeding and variety testing
Piepho, H. P., Möhring, J., Melchinger, A. E., & Büchse, A. (2008) · Euphytica · doi:10.1007/s10681-007-9449-8 · Cited by 862 · cited 8×
“Fourth, our main result applies to BLUP as well as to BLUE of linear functions of interest, while Mathew and Nordström only consider BLUE.”
BackgroundMethodInfluentialSelf-citationDeregressing estimated breeding values and weighting information for genomic regression analyses
Garrick, D. J., Taylor, J. F., & Fernando, R. L. (2009) · Genetics Selection Evolution · doi:10.1186/1297-9686-41-55 · Cited by 653
Smith, A., Cullis, B., & Thompson, R. (2001) · Biometrics · doi:10.1111/j.0006-341x.2001.01138.x · Cited by 566 · cited 10×
“For example, Smith et al. (2001a,2001b) obtain environment-specific BLUPs of genotype performances using a model with random genotypes and fixed environments.”
BackgroundMethodInfluentialOgutu, J. O., Schulz-Streeck, T., & Piepho, H.-P. (2012) · BMC Proceedings · doi:10.1186/1753-6561-6-s2-s10 · Cited by 440 · cited 1×
“When using some machine learning method that implies a linear regression on markers (Ogutu et al., 2012), genotypic main effects can be captured in βa, while ua can be omitted.”
MethodSelf-citationStatistical Meta‐Analysis with Applications
Hartung, J., Knapp, G., & Sinha, B. K. (2008) · Wiley Series in Probability and Statistics · doi:10.1002/9780470386347 · Cited by 418 · cited 2×
“This kind of heterogeneity is the rule rather than the exception in plant breeding and cultivar testing, and it is also very commonly encountered in individual patient data (Hartung et al., 2008).”
BackgroundResultPrediction of Maize Single‐Cross Performance Using RFLPs and Information from Related Hybrids
Bernardo, R. (1994) · Crop Science · doi:10.2135/cropsci1994.0011183x003400010003x · Cited by 379 · cited 1×
“For breeding value estimation (Bernardo, 1994), genomic selection (Bernardo and Yu, 2007; Piepho, 2009) or association mapping (Yu et al., 2006; Müller et al., 2011), it is convenient to first summarise the phenotypic data across environments by computing adjusted means for each genotype, and then…”
BackgroundNelder, J. A. (1965) · Proceedings of the Royal Society of London. Series A. Mathematical and Physical Sciences · doi:10.1098/rspa.1965.0012 · Cited by 305 · cited 1×
“These factors, which define the hierarchy, may be referred to as block factors (Nelder, 1965) and comprise all factors that are innate to the experimental units (Brien, 1983).”
BackgroundRidge Regression and Extensions for Genomewide Selection in Maize
Piepho, H. P. (2009) · Crop Science · doi:10.2135/cropsci2008.10.0595 · Cited by 271 · cited 7×
“When fitting a random effects model to the genetic effect, βa will just be a common intercept, while genotypic main effects are modelled via ua, for which a ridge regression model or a spatial model can be fitted in stage three (Piepho, 2009).”
BackgroundMethodInfluentialSelf-citationAnalyzing genotype-environment data by mixed models with multiplicative terms
Piepho, H.-P. (1997) · Biometrics · doi:10.2307/2533976 · Cited by 234 · cited 1×
“…interaction effects (GE), for example, when producing the so-called GGE biplots (Yan and Kang, 2003), or when fitting factor-analytic models (Piepho, 1997) to obtain environment-specific BLUPs that borrow strength from across environments (Smith et al., 2001a,2001b; Welham et al., 2010).”
Self-citationComparison of Weighting in Two‐Stage Analysis of Plant Breeding Trials
Möhring, J., & Piepho, H. ‐. (2009) · Crop Science · doi:10.2135/cropsci2009.02.0083 · Cited by 228 · cited 5×
“In contrast, a two-stage approach where each environment is analysed individually in the first stage has great practical appeal because of both its simplicity and computational efficiency (Möhring and Piepho, 2009).”
BackgroundMethodInfluentialSelf-citationApplications: The Analysis of Crop Variety Evaluation Data in Australia
Smith, A., Cullis, B., & Gilmour, A. (2001) · Australian & New Zealand Journal of Statistics · doi:10.1111/1467-842x.00163 · Cited by 217 · cited 10×
“For example, Smith et al. (2001a,2001b) obtain environment-specific BLUPs of genotype performances using a model with random genotypes and fixed environments.”
BackgroundMethodInfluentialvan Eeuwijk, F. A., Kang, M., & Denis, J. (1996) · CRC Press eBooks · doi:10.1201/9781420049374.ch2 · Cited by 177 · cited 2×
“…released crop cultivars will be grown in agroecologically diverse target regions, it is important to test candidate genotypes in a representative sample of environments from the target region and to assess genotype–environment interaction, which is typically substantial (van Eeuwijk et al., 1996).”
BackgroundThompson, R., Cullis, B., Smith, A., & Gilmour, A. (2003) · Australian & New Zealand Journal of Statistics · doi:10.1111/1467-842x.00297 · Cited by 113 · cited 1×
“The rotation of means yields non-sparse design matrices for fixed and random effects, which limits the use of sparse matrix methods (Thompson et al., 2003), while the design matrices for the unrotated means may be of a form suited to the use of such methods.”
BackgroundBest Linear Unbiased Prediction of Cultivar Effects for Subdivided Target Regions
Piepho, H., & Möhring, J. (2005) · Crop Science · doi:10.2135/cropsci2004.0398 · Cited by 101 · cited 1×
“For this purpose, one would fit a model with fixed sub-regions, but random environments within regions (Piepho and Möhring, 2005).”
BackgroundSelf-citationA comparison of analysis methods for late-stage variety evaluation trials
Welham, S. J., Gogel, B. J., Smith, A. B., Thompson, R., & Cullis, B. R. (2010) · Australian & New Zealand Journal of Statistics · doi:10.1111/j.1467-842x.2010.00570.x · Cited by 98 · cited 4×
“Optimal performance of this approach has been substantiated based on simulation evidence (Welham et al., 2010).”
BackgroundMethodInfluentialEvaluation of experimental designs and spatial analyses in wheat breeding trials
Qiao, C. G., Basford, K. E., DeLacy, I. H., & Cooper, M. (2000) · Theoretical and Applied Genetics · doi:10.1007/s001220050002 · Cited by 91 · cited 1×
“For example, some environments may require spatial modelling, while others may best be analysed by a randomisation-based model (Qiao et al., 2000; Piepho et al., 2008b).”
BackgroundModelling expectation and variance for genotype by environment data
Denis, J., Piepho, H., & van Eeuwijk, F. A. (1997) · Heredity · doi:10.1038/hdy.1997.139 · Cited by 89 · cited 1×
“We assume that the tested environments can be considered as a random sample from a target population of environments (Denis et al., 1997).”
BackgroundSelf-citationAnalysis of series of variety trials
Patterson, H. D. (1997) · Springer eBooks · doi:10.1007/978-94-009-1503-9_9 · Cited by 78 · cited 1×
“In (20) fixed effects appear before and random effects after the colon operator (:), and the dot operator (•) denotes crossing of two factors (Patterson, 1997).”
BackgroundSauer, N., Emrich, K., Piepho, H., Lemme, A., Redshaw, M., & Mosenthin, R. (2008) · Poultry Science · doi:10.3382/ps.2007-00514 · Cited by 75 · cited 1×
“Second, our approach also works when the treatment–trial classification is not complete, i.e. when some treatments are not tested in some trials, a situation that is quite common in plant breeding and also in other fields such as animal nutrition (Sauer et al., 2008).”
BackgroundSelf-citationNearest Neighbour Adjustment and Linear Variance Models in Plant Breeding Trials
Piepho, H., Richter, C., & Williams, E. R. (2008) · Biometrical Journal · doi:10.1002/bimj.200710414 · Cited by 70 · cited 6×
“Fourth, our main result applies to BLUP as well as to BLUE of linear functions of interest, while Mathew and Nordström only consider BLUE.”
BackgroundMethodInfluentialSelf-citationStatistical Learning from a Regression Perspective
Maindonald, J. (2009) · Journal of Statistical Software · doi:10.18637/jss.v029.b12 · Cited by 68
Comparison of One‐Step and Two‐Step Meta‐Analysis Models Using Individual Patient Data
Mathew, T., & Nordström, K. (2010) · Biometrical Journal · doi:10.1002/bimj.200900143 · Cited by 55 · cited 3×
“In the discussion, we compare our main result to closely related results derived recently in the context of meta-analysis (Mathew and Nordström, 2010).”
ResultAnalysis of Variance Tables Based on Experimental Structure
Brien, C. J. (1983) · Biometrics · doi:10.2307/2530806 · Cited by 43 · cited 1×
“These factors, which define the hierarchy, may be referred to as block factors (Nelder, 1965) and comprise all factors that are innate to the experimental units (Brien, 1983).”
BackgroundMüller, B. U., Stich, B., & Piepho, H. (2010) · Heredity · doi:10.1038/hdy.2010.125 · Cited by 42 · cited 1×
“For breeding value estimation (Bernardo, 1994), genomic selection (Bernardo and Yu, 2007; Piepho, 2009) or association mapping (Yu et al., 2006; Müller et al., 2011), it is convenient to first summarise the phenotypic data across environments by computing adjusted means for each genotype, and then…”
BackgroundSelf-citationOn the use of multiple lattice designs and α‐designs in plant breeding trials
Piepho, H., Büchse, A., & Truberg, B. (2006) · Plant Breeding · doi:10.1111/j.1439-0523.2006.01267.x · Cited by 40 · cited 1×
“We note in passing that while this type of design with multiple lattice trials per environment is very commonly used by German plant breeding companies, its use should be discouraged because of poor statistical efficiency (Piepho et al., 2006).”
BackgroundSelf-citation
Research works printed but not linked: 1
- Rao C. R., 2008, Least Squares and Alternatives
Cited by
van Eeuwijk, F. A., Bustos‐Korts, D. V., & Malosetti, M. (2016) · Crop Science · doi:10.2135/cropsci2015.06.0375 · Cited by 286
Heritability in Plant Breeding on a Genotype-Difference Basis
Schmidt, P., Hartung, J., Bennewitz, J., & Piepho, H.-P. (2019) · Genetics · doi:10.1534/genetics.119.302134 · Cited by 214
“SAS Institute Inc., Cary, NC. Schmidt, P., J. Hartung, J. Rath, and H.-P. Piepho, 2019 Estimating broad-sense heritability with unbalanced data from agricultural cultivar trials.”
InfluentialSelf-citationMarker-Based Estimation of Heritability in Immortal Populations
Kruijer, W., Boer, M. P., Malosetti, M., Flood, P. J., Engel, B., Kooke, R., Keurentjes, J. J. B., & van Eeuwijk, F. A. (2015) · Genetics · doi:10.1534/genetics.114.167916 · Cited by 198
Smith, A. B., Ganesalingam, A., Kuchel, H., & Cullis, B. R. (2015) · Theoretical and Applied Genetics · doi:10.1007/s00122-014-2412-x · Cited by 172
“The analysis of long-term MET data from these programs (both state-based CVT and more recently NVT) has followed the same approach since the mid 1990s, and is documented in Smith et al. (2001a). This involved the use of a variance component model with V × E interaction partitioned into sources associated with pre-defi…”
MethodResultInfluentialEstimating Broad‐Sense Heritability with Unbalanced Data from Agricultural Cultivar Trials
Schmidt, P., Hartung, J., Rath, J., & Piepho, H. (2019) · Crop Science · doi:10.2135/cropsci2018.06.0376 · Cited by 143
Self-citation
Metadata
- Authors
- Piepho, H., Möhring, J., Schulz‐Streeck, T., & Ogutu, J. O.
- Affiliations
- University of Hohenheim (DE)
- Corresponding author
- Piepho, Hans‐Peter
- Venue
- Biometrical Journal
- Publisher
- Wiley
- Published
- 2012-09-25
- Volume
- 54
- Issue
- 6
- Pages
- 844-860
- ISSN
- 0323-3847 (ISSN-L)
- 1521-4036
- OpenAlex
- W1602214942
- Cited by
- 193
- References
- 30
- Citations by year
- 2026: 23 · 2025: 14 · 2024: 17 · 2023: 14 · 2022: 18 · 2021: 17 · 2020: 13 · 2019: 12 · 2018: 11 · 2017: 9 · 2016: 10 · 2015: 11 · 2014: 15 · 2013: 7 · 2012: 2
- FWCI
- 12.80
- Type
- journal article
- Language
- en
- Open access
- closed
- Locations
- Biometrical Journal (publishedVersion)
- PubMed (publishedVersion)
- Topics
- Life Sciences › Agricultural and Biological Sciences › Plant Science › Genetics and Plant Breeding
- Life Sciences › Agricultural and Biological Sciences › Agronomy and Crop Science › Agriculture, Soil, Plant Science
- Social Sciences › Decision Sciences › Management Science and Operations Research › Optimal Experimental Design Methods
- Keywords
- multi-environment trials, two-stage analysis, mixed models, covariance matrix, adjusted means, diagonalization, linear operators, meta-analysis, gene-environment interaction
- MeSH terms
- Analysis of Variance, Biometry, Breeding, Chromosome Mapping, Environment, Genotype, Humans, Plants, Models, Statistical, Genome, Plant, Quantitative Trait Loci
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.