International Adaptation Trial: Using probe and reference genotypes to characterize global spring wheat production environments.
1The School of Land and Food Sciences, The University of Queensland, St Lucia QLD 4072 Australia Email kmathews@uq.edu.au
2CSIRO Plant Industry, Queensland Biosciences Precinct, 306 Carmody Rd St Lucia QLD 4067 Australia Email Scott.Chapman@csiro.au
3CIMMYT, Apdo. Postal 6-641, 06600 Mexico, D.F. Mexico, www.cimmyt.cgiar.org
Results are presented from a nursery (International Adaptation Trial - IAT) to investigate environmental stresses in spring-wheat production areas. These illustrate concepts of using probe and reference genotype sets to characterize environments. A probe pair compares presence/absence of the Rht1 dwarfing gene in a Nesser background with results presented on maps. A reference genotype set of twenty-one broadly adapted CIMMYT-derived lines, grouped seventy-eight trials into three main groups, Australian rain-fed, high-yielding irrigated international trials and lower yielding irrigated or rain-fed international trials.
Experiments in global and Australian locations aims to assist Australian wheat breeders to select germplasm for improved yield in our wheat industry.
Key Words
Genotype-by-environment interaction, probe genotypes, reference genotypes
Genotype-by-environment interaction (GEI) complicates the interpretation of multi-environment trials (METs) in plant breeding, (e.g. Peterson and Pfeiffer, 1989; DeLacy et al., 1996). After estimating GEI effects, the next step is to characterize where trials are grown in an attempt to identify the repeatable sources of genotype-by-environment variation. Breeding strategies can then be designed to account for or to exploit GEI. Direct methods to characterize environments use data on the climate, trial management practices and soil parameters. Cooper and Fox (1996) discuss the use of ‘probe’ and ‘reference’ genotype sets to indirectly characterize environments. Probe sets aim to reveal specific environmental challenges (e.g. a disease or soil chemical problem), whereas reference sets are random-effects bioassays to investigate the relationships among environments. The International Adaptation Trial (IAT) is an investigative spring wheat nursery distributed globally by the International Centre for Wheat and Maize Improvement (CIMMYT). It contains probe genotypes to assess biotic and abiotic stress and a reference set of broadly adapted germplasm from CIMMYT and Australia to determine relationships among Australian and CIMMYT breeding and testing locations. The concepts of both probe and reference genotype sets in characterising global spring-wheat environments are illustrated using the IAT.
The IAT contains 60 bread and 20 durum wheat lines, primarily of CIMMYT and Australian origin, chosen for their drought adaptation characteristics and, their ability to identify soil borne problems (abiotic and biotic) and agronomic traits (Table 1). In many cases there are isogenic pairs (usually derived by backcrossing) available for trait comparisons. Where an isogenic pair does not exist, the contrast consists of lines with similar genetic backgrounds but known different reactions for a key trait. There are about 40 contrasts, including ‘replicate’ contrasts for several traits.
Table 1. Trait contrasts in the International Adaptation Trial using probe and reference genotype sets.
Agronomic |
Soil Constraints |
Disease |
Adaptive |
Plant height (Rht1 or 2) |
Boron toxicity |
Crown rot |
Terminal drought (1ME4A) |
Vernalisation/earliness |
Zn deficiency |
Common root rot |
Pre-anthesis drought (ME4B) |
Drought adaptation |
Mn deficiency |
Cereal cyst nematode |
Residual moisture (ME4C) |
Acid soil |
Root lesion nematode |
Irrigated environments (ME1) | |
Major rust genes |
High rainfall (ME2) |
1 ME = Mega-environment classification as defined by CIMMYT (Calhoun et al., 1994)
Twenty-one CIMMYT-derived lines are used here to investigate relationships among environments. These lines were included for their drought adaptation to spring wheat growing environments or their broad adaptation to both drought and irrigated or high rainfall conditions (Table 2).
Table 2. Broadly adapted CIMMYT-derived genotypes with trait descriptions.
Code |
Genotype Name |
Trait Description |
ATTL |
ATTILA |
Adaptation to drought & irrigation conditions (ME4 & 1) (India) |
CETT |
CETTIA |
Global adaptation to drought (ME4) |
CHIL |
CHIL/PRL |
Global adaptation to drought (ME4) |
CNDO |
CNDO/R143//ENTE/MEXI |
Synthetic derivative with good drought tolerance in Mexico |
DHAR |
DHARWAR DRY |
Adaptation in monsoonal India (Rajastan) |
GLVZ |
GALVEZ DWARF |
Rht1 isoline - long term drought check at CIMMYT |
HXL |
HXL7573/2*BAU |
Adaptation to drought tolerance (Heilongjiang, north east China) |
INQA |
INQALAB 91 |
World’s greatest acreage; early maturing (Pakistan) |
JUN |
JUN/BOMB |
Global adaptation to drought (ME4) |
KAUZ |
KAUZ DWARF |
Rht1 isoline - released in South Asia - high yield potential |
NESS |
NESSER DWARF |
Rht1 isoline – widely released in West Asia |
PAVN |
PAVON DWARF |
Rht2 isoline – widely released for dry and irrigated systems |
PFDL |
PROINTA FEDERAL |
Bobwhite line - tolerant to pre-anthesis stress (Argentina) |
PRL |
PRL/SARA//TSI/VEE#5 |
Hessian fly resistant |
PSTO |
PASTOR*2/OPATA |
Global adaptation to drought & high rainfall (ME4 & 2) |
PSTR |
PASTOR |
Global adaptation to drought, high rain & irrigation (ME4, 2 & 1) |
SITT |
SITTA |
Global adaptation to drought conditions (ME4) |
SON |
SONALIKA |
Very early maturing; Green Revolution variety in South Asia |
SSER |
SUPER SERI #1 |
SERI plus lr19 gene - widely adapted and released |
TUI |
TUI |
Global adaptation to drought & high rainfall (ME4 & 2) |
URES |
URES/JUN//KAUZ |
Global adaptation to drought & irrigation (ME4 & 1) |
In most cases, the bread and durum lines were grown in separate two-replicate α-lattice designs under local agronomic practice. Where fungicide was not applied, trials that either reported a severe score for foliar disease or had a significant contrast for leaf or stem rust were removed. The remaining 78 trials were sown between 2000 and 2003 (25 were in Australia, 14 at CIMMYT’s research station northwestern Mexico and the rest in major spring wheat production areas) (Fig. 1). Assuming replicate, replicate x block and variety effects to be random, trait contrasts were made to test the importance of the presence/absence of each trait (SAS Proc Mixed). Best linear unbiased predictors (BLUPs) were calculated for each trial and averaged for lines representing the presence or absence of the trait. Thus, “contrast averages”, were mapped on the basis of statistical significance, (P < 0.10).
Extended factor analytic techniques (Smith et al. 2001) were applied in the across site analysis using ASREML (Gilmour et al 2002). Factor analytic models are the random effect equivalent of the fixed AMMI (additive main effects and multiplicative interactions) models and allow fitting of separate genetic variances to each environment and genetic covariances among environments. Best spatial models (i.e. row and column effects) where possible, otherwise incomplete block analysis was used with replicate, replicate × block and variety considered as random effects. Design terms with a variance component of zero were removed to achieve a more parsimonious model. A biplot created from the factor analytic loadings and scores assists interpretation of the genetic correlations among environments. The circle in a biplot indicates 100% of trial genetic variance explained, i.e. trials whose vector length equals the radius of the circle have 100% of their genetic variance explained by the two factors on this biplot (Smith, A., 2004, pers comm.)
An example map for Rht genes (Fig. 1) shows that the yield difference between semi-dwarf and tall types in a Nesser background is generally not significant. In one trial the tall genotype yields more than the semi-dwarf and in high input trials in Mexico, several trials in Australia and trials in Pakistan, Argentina, Spain and Ecuador, the semi-dwarf out yielded the tall type. The green revolution was based on the yield advantage of semi-dwarf types when combined with optimum agronomy and inputs. This preliminary analysis shows that following substantial improvement of breeding in a semi-dwarf backgrounds, the re-introduction of the ‘tall’ allele does not have as large a negative impact as might be expected, i.e. in the absence of lodging, the breeders were able to make significant progress for other adaptive traits. In a smaller set of trials, Singh et al (2001) noted that the value of the dwarfing gene in a large set of backgrounds was 0.66 t/ha. Further analysis of our data will investigate this effect in a large number of locations and in several genetic backgrounds.
Figure 1. Distribution of International Adaptation Trial, 2000-2003. The contrast shown is for the Nesser near-isogenic Rht1 pair.
The centre point of the biplot approximates ‘average’ yield in all environments, while the cosine of the angle between any two vectors is the genetic correlation between the two environments. Hence, in this dataset relationships between pairs of environments span the full range of highly correlated (coincident lines) to uncorrelated (angle = 90°) to negatively correlated (90 < angle < 180°). In the vertical direction, factor 1 was correlated with average yield in each environment ( r = 0.47).
The centre group (partially correlated with Australian environments) contains the higher yielding southern European trials and the irrigated trials from CIMMYT. Most of the environments in the largely irrigated Indo-Gangetic plains and the non- or partially irrigated trials of CIMMYT are grouped toward the left hand side. The strong correlation between Obregon and the rice-wheat environments of South Asia confirms CIMMYT’s historical success in developing spring wheat germplasm that is well adapted to this huge area of irrigated wheat production. However, these environments to the left were poorly correlated with the rain-fed Australian environments. The exceptions were: 1) two high-yielding Australian sites (Tamworth and Willowtree) located in the irrigated group; and 2) an irrigated CIMMYT trial, C149, that was grown in irrigation on melgas, a planting system similar to the Australian flat system with which it clusters.
Across the dataset, Inqalab has the broadest adaptation, being toward the centre upper part of the biplot, i.e. except for environments A4 and A11, this cultivar has a positive score when a perpendicular projection is made on to any environment vector. Nesser is the most poorly adapted in broad terms. Tui, Prointa Federal and the early maturing Sonalika have broad adaptation to international environments while Dharwar Dry, Attila and Kauz are the most broadly adapted to Australian environments.
Figure 2. Twenty-one broadly adapted CIMMYT-derived lines showing relationships among 78 global environments, variance explained is 56%. The genotypes are labelled using the Code in Table 2; “A” = Australian environment, “C”= CIANO –Obregon, CIMMYT breeding location, “I”= International spring wheat location. The circle represents 100% explained genetic variance.
Probe and reference genotype sets are useful to indirectly characterise environments, thereby explaining repeatable genotype-by-environment interaction. In conjunction with maps, probe genotypes assist breeders to identify locations or regions which allow the phenotypic expression of specific adaptation for a particular trait. Reference genotype sets allow a broader interrogation of the relationships among environments. These 21 lines classified the environments into three loosely geographical groups: rain fed Australian, irrigated high yielding CIMMYT and southern European locations and the non-irrigated CIMMYT and irrigated South Asian locations. A more direct result for Australian breeders is that CIMMYT international nurseries where Attila and Kauz perform well could be a source of new germplasm for yield adaptation in Australia.
We would like to thank the many wheat breeders who ran these trials throughout the world and the GRDC for funding this research.
Calhoun DS, Gebeyehu G, Miranda A, Rajaram S and van Ginkel M (1994) Choosing evaluation environments to increase wheat grain yield under drought conditions. Crop Science 34, 673-678
Cooper M and Fox PN (1996) Environmental characterization based on probe and reference genotypes. In ‘Plant Adaptation and Crop Improvement’ (Eds. M Cooper and GL Hammer) pp. 529-547 (CAB International)
DeLacy IH, Basford KE, Cooper M, Bull JK, McLaren CG (1996) Analysis of multi-environment trials – an historical perspective. In ‘Plant Adaptation and Crop Improvement’ (Eds. M Cooper and GL Hammer) pp. 39-124 (CAB International)
Gilmour AR, Gogel BJ, Cullis BR, Welham SJ and Thompson R (2002) ASReml User Guide Release 1.0 VSN International Ltd, Hemel Hempstead, HP11ES, UK
Peterson CJ and Pfeiffer WH (1989) International winter wheat evaluation: relationships among test sites based on cultivar performance. Crop Science 29, 276-282
Singh RP, Huerta-Espino J, Rajaram S and Crossa J (2001) Grain yield and other traits of tall and dwarf isolines of modern bread and durum wheats. Euphytica 119, 241–244
Smith A, Cullis BR, Thomson R (2001) Analyzing variety by environment data using multiplicative mixed models and adjustments for spatial field trend. Biometrics 57, 1138-1147