RESEARCH

Genetic Diversity of a Maize Association

Population with Restricted Phenology
Candice N. Hansey, James M. Johnson, Rajandeep S. Sekhon, Shawn M. Kaeppler, and Natalia de Leon*

Dep. of Agronomy, Univ. of Wisconsin-Madison, Madison, WI 53706.

ABSTRACT Received 29 Mar. 2010. *Corresponding author (ndeleongatti@wisc.edu).
Association analysis to identify genes and
Abbreviations: BSSS, Iowa Stiff Stalk Synthetic; GDD, growing degree
alleles underlying quantitative genetic variation
days; GEM, Germplasm Enhancement of Maize; LD, linkage disequilib-
is growing in value as the density of genotypic
rium; MDS, multidimensional scaling; NAM, nested association mapping;
data increases. Association panels are often
NC7, North Central Regional Plant Introduction Station; NSS, Non-Stiff
chosen to maximize molecular and pheno-
Stalk Synthetic; PIC, polymorphism information content; QTL, quan-
typic diversity, but traits measured at harvest
titative trait loci; SNP, single nucleotide polymorphism; SSS, Stiff Stalk
require restricted phenology to obtain data on
Synthetic; UPGMA, unweighted pair group method with arithmetic mean.
plants that mature within a comparable interval.
The objective of this study was to characterize
a set of inbred lines that will mature reliably in
the upper midwestern United States for the pur-
pose of assessing traits relevant to grain and
G enetic dissection of quantitative traits is important for
improving the efficiency of breeding programs and as a dis-
covery tool to understand basic processes in plant development,
stover yield as well as stover quality. A total of physiology, and biochemistry. Linkage mapping has traditionally
1411 lines from the North Central Regional Plant been used to identify quantitative trait loci (QTL) underlying phe-
Introduction Station, our program, and contrib-
notypic variation. Recently the use of linkage disequilibrium (LD)
uted by collaborators were grown in observation
mapping to dissect quantitative traits has dramatically increased.
plots. A set of 627 lines were chosen based on
flowering within the desired interval, produc-
Linkage disequilibrium mapping may have an advantage over tra-
tion of viable seed, agronomic suitability, uni- ditional linkage mapping in increased resolution depending on the
formity, and pedigree information. Flowering species and population and has the ability to evaluate an array of
time ranged from 944 to 1645 growing degree alleles in diverse germplasm without the need for structured popu-
days (GDD). Genotypic data for the 627 lines lations. Elite breeding materials of many crop species have limited
was obtained using a 1536 Illumina Golden- diversity due to domestication and selection bottlenecks (Yama-
Gate assay. The panel offers deep replication saki et al., 2007). To maximize assessment of allelic diversity, asso-
of most Midwest dent alleles. There is a lesser ciation mapping panels frequently utilize germplasm outside of
representation of tropical alleles relative to other these elite materials. Knowledge of alleles from diverse material
association panels but includes some Germ- associated with important agronomic traits is useful in the pro-
plasm Enhancement of Maize (GEM)-derived
cess of developing improved germplasm. However, as genotypic
lines and early flowering lines of tropical origin.
data availability increases the distinction between linkage (fam-
The information described in this manuscript is
anticipated to facilitate the use of this material
ily) and LD (population) mapping will dissolve and high quality
for association genetic studies.
Published in Crop Sci. 51:704–715 (2011).
doi: 10.2135/cropsci2010.03.0178
Published online 10 Jan. 2011.
© Crop Science Society of America | 5585 Guilford Rd., Madison, WI 53711 USA
All rights reserved. No part of this periodical may be reproduced or transmitted in any
form or by any means, electronic or mechanical, including photocopying, recording,
or any information storage and retrieval system, without permission in writing from
the publisher. Permission for printing and for reprinting the material contained herein
has been obtained by the publisher.

704 WWW.CROPS.ORG CROP SCIENCE, VOL. 51, MARCH– APRIL 2011

phenotypic data on relevant germplasm will be the more (Camus-Kulandaivelu et al., 2006; Ducrocq et al., 2008,
critical considerations (Myles et al., 2009). Technological 2009). A set of 632 temperate, tropical, and subtropical inbred
advances such as next generation sequencing are making it lines, and a core set of 60 lines representing 90% of the haplo-
possible to obtain large genotypic data sets, which facilitate type diversity of the larger set, has been described (Yan et al.,
high-resolution mapping across diverse genotypes. 2009). Wang et al. (2008b) characterized the genetic diver-
Association mapping has been used in various plant sity and LD for a set of 95 diverse Chinese inbred lines. In
species including barley (Hordeum vulgare L.) (Kraakman et another report, Xie et al. (2008) described population struc-
al., 2006; Stracke et al., 2009), soybean (Glycine max (L.) ture, kinship, and LD for a set of 187 Chinese maize inbred
Merr.) (Jun et al., 2008; Wang et al., 2008a), wheat (Triticum lines. Genetic diversity of CIMMYT maize inbred lines and
aestivum L.) (Breseghello and Sorrells, 2006; Crossa et al., open pollinated populations have also been characterized and
2007), sorghum (Sorghum bicolor (L.) Moench) (Casa et al., have been incorporated into diversity panels (Warburton et
2008, Murray et al., 2009), rice (Oryza sativa L.) (Olsen and al., 2002; Reif et al., 2004). While these resources capture
Purugganan, 2002; Bao et al., 2006; Agrama et al., 2007), genetic variation in maize, adaptation and phenology of the
sugarcane (Saccharum officinarum L.) (Wei et al., 2006), and germplasm was not taken into account. For example, in tem-
Arabidopsis thaliana (L.) Heynh. (Olsen et al., 2004; Aran- perate regions, a significant portion of these diversity panels
zana et al., 2005; Ehrenreich et al., 2007; Zhao et al., 2007). will not flower and mature in their testing area thereby limit-
Many successful examples also exist for maize (Zea mays L.). ing the use of these panels. A diversity panel with restricted
The first quantitative trait dissected using association map- phenology that maintains genetic and phenotypic diversity
ping in maize was flowering time (Thornsberry et al., 2001; will allow relevant assessment of phenotypes in short-season
Andersen et al., 2005; Camus-Kulandaivelu et al., 2006; environments. Our objectives were to (i) identify a set of
Salvi et al., 2007; Ducrocq et al., 2008). Beló et al. (2008) diverse inbred genotypes that will mature in the upper Mid-
used whole genome association mapping to identify associa- west region of the United States, (ii) evaluate the genetic and
tion between fatty acid desaturase 2 ( fad2) and increased oleic phenotypic diversity in the diversity panel, (iii) determine
acid levels in maize. Wilson et al. (2004) assessed the associ- population substructure and familial relationships for use in
ation of genes in the starch biosynthetic pathway with ker- association mapping, (iv) compare the relationship between
nel composition and starch pasting properties by evaluating pedigree and genetic relatedness, and (v) demonstrate the
candidate genes for their association with phenotype. Pres- utility of this population for association mapping using the
soir et al. (2009) employed a combined approach of link- previously described Vgt1 loci as a proof of concept.
age analysis and LD methods to determine the importance
of barren inflorescence2 (bif2) in flowering time as well as a MATERIALS AND METHODS
wide range of maize architecture traits. Harjes et al. (2008) Germplasm
identified the effect of alleles at the lycopene epsilon cyclase A total of 1164 inbred lines were provided by Mark Millard at the
(lcyE) locus on carotenoid accumulation in the endosperm North Central Regional Plant Introduction Station (NC7; Iowa
using a combined approach of linkage mapping and associa- State University, Ames, IA) based on information that they would
tion analysis. Szalma et al. (2005) showed that anthocyanin- be likely to flower in 105 d or less. In addition, inbred lines were pro-
less1 (a1) and white pollen1 (whp1) are associated with maysin vided by University of Guelph (Dr. Elizabeth Lee, Department of
synthesis in maize silks and identified epistatic interactions Plant Agriculture, University of Guelph, Guelph, ON), University
among the alleles. Zheng et al. (2008) further dissected the of Delaware (Dr. James Hawk, Plant and Soil Sciences, University
molecular basis of previously identified QTL qHO6 and of Delaware, Newark, DE), Iowa State University (Committee for
demonstrated the association of this QTL to natural varia- Agricultural Development and Iowa Crop Improvement Associa-
tion, Iowa State University, Ames, IA), University of Pennsylvania
tion in kernel oil and oleic acid content. This extensive list
(John Shaffer, Department of Crop and Soil Sciences, Pennsylvania
demonstrates that mining of the genetic diversity in maize State University, University Park, PA), North Carolina State Uni-
can be accomplished via linkage mapping and association versity (Dr. Matthew Krakowsky, ARS Plant Science Research,
analysis for many phenotypic traits using different sets of North Carolina State University, Raleigh, NC), and developmental
genetic materials and various statistical approaches. lines from our program at the University of Wisconsin (Dr. Nata-
There are well-described, publicly available resources for lia de Leon, Department of Agronomy, University of Wisconsin,
association mapping in maize. Phenotypic and genetic diver- Madison, WI) (Supplemental Table 1). Phenotypic assessment of
sity of a diverse set of 302 maize inbred lines was character- these 1411 inbred lines was performed during the summer of 2008
ized (Liu et al., 2003; Flint-Garcia et al., 2005). Later, 25 at the West Madison Agricultural Research Station in Madison,
lines from this set were used to develop a nested association WI. Highly related lines (e.g., single gene introgression lines) were
mapping (NAM) population aimed at maximizing diversity generally removed. A subset of 627 of the lines (hereafter referred
(McMullen et al., 2009). A set of 375 diverse inbred lines to as the Wisconsin diversity set) was selected based on flowering
within the desired interval, sufficient vigor for trialing, uniformity,
representing American and European diversity was described
and pedigree information. A total of 216 lines from the Wisconsin
and has been used in subsequent association mapping studies diversity set were in common with the Flint-Garcia et al. (2005)

CROP SCIENCE, VOL. 51, MARCH– APRIL 2011 WWW.CROPS.ORG 705

diversity set (hereafter referred to as the Goodman-Buckler diver- probability that two alleles randomly chosen from a population
sity set) (Supplemental Table 1). are different. Gene diversity was calculated at each locus as:
k

∑ P
2
D̂ = (1 – u )/[1 – (1 + f )/n],
Pedigree u =1

Pedigree information was compiled from multiple sources (Gerdes in which Pu is the frequency of the uth allele, n is the sample
et al., 1993; Liu et al., 2003; Flint-Garcia et al., 2005; Mikel, size, and f is the inbreeding coefficient estimated from genotype
2006; Mikel and Dudley, 2006; USDA, ARS, National Genetic frequencies (Weir, 1996). Polymorphism information content
Resources Program [available at http://www.ars-grin.gov/cgi-bin/ is another related measure of genetic diversity; for a given loci,
npgs/html/tax_site_acc.pl?NC7%20Zea%20mays%20subsp.%20 PIC was calculated as:
mays; verified 6 Dec. 2010]) and restructured in a previously pro- k k −1 k

∑ P ∑ ∑
2 2 2
n =1–
PIC – 2P uP v ,
posed standard pedigree format (Purdy et al., 1968). Coefficients of u =1
u
u = 1v = u + 1
coancestry were calculated using the computer program RELATE
in which for genotype AuAv, Pu is the frequency of the uth allele
(Bernardo et al., 1997) to produce a relatedness matrix of all inbred
at Au and P v is the frequency of the vth allele at Av (Botstein et
lines (Emik and Terrill, 1949). This program utilizes a modifica-
al., 1980).
tion of the tabular method to calculate the coefficients. The tabular
These summary statistics were calculated for the entire diver-
analysis procedure was modified to account for full inbreeding of
sity panel as well as specific subsets of the panel. The so-called “cate-
individuals and unequal parental contributions. For lines derived
gory” subsets included genotypes unique to the Wisconsin diversity
from sources of unknown origin, the parental contribution from
set, genotypes from the Goodman-Buckler diversity set that meet
that source was considered zero. This will impact germplasm
our phenology restrictions, genotypes from the Goodman-Buckler
derived outside of the United States more than U.S. germplasm
(Supplemental Table 1). Lines derived from open pollinated popu- diversity set that do not meet the phenology restrictions, and all
lations were assumed unrelated in the analysis due to the lack of genotypes from the Goodman-Buckler diversity set. The “maturity
sufficient information regarding the degree of relatedness. group” subsets were determined based on average flowering time
measurements from 2008 and 2009. Maturity groups correspond
to classic industry groupings in days after planting: time to flow-
Molecular Marker Genotyping ering is less than 75 d after planting (2008: 1210 growing degree
Of the 627 lines included in this study, 411 were genotyped days [GDD]; 2009: 1046 GDD), 75 to 85 d after planting (2008:
using the Illumina GoldenGate high-throughput single nucleo- 1210–1430 GDD; 2009: 1046–1214 GDD), 85 to 95 d after plant-
tide polymorphism (SNP) assay (Fan et. al., 2006). Seedling leaf ing (2008: 1430–1601 GDD; 2009: 1214–1410 GDD), 95 to 105
tissue from five to ten plants was bulked for DNA extraction. d after planting (2008: 1601–1789 GDD; 2009: 1410–1607 GDD),
DNA was extracted using the cetyl(trimethyl)ammonium bro- and greater than 105 d after planting (2008: 1789 GDD; 2009: 1607
mide (CTAB) method (Saghai-Maroof et al., 1984). A set of 1536 GDD). Population substructure groups were assigned as the sub-
SNP marker data points was obtained for each genotype. The structure group with the majority membership for each genotype.
remaining 216 lines had already been genotyped using the same If a genotype did not have greater than 50% majority membership
Illumina GoldenGate high-throughput assay (McMullen et al., in any group it was assigned to a “mixed group” (group 9). This
2009). Genotypic data of these inbred lines was obtained from membership threshold was used simply for illustration purposes.
the website (http://www.panzea.org/lit/data_sets.html#genos PowerMarker was also used to calculate genetic relatedness
[verified 6 Dec. 2010]). A subset of the 1536 SNPs were origi- as pair-wise Rogers distances (Rogers, 1972) based on the 511
nally designed to have a unique allele in inbred line B73 for map- unbiased SNP markers. Rogers distances were calculated as:
ping the NAM populations, which resulted in a bias toward B73. m aj

The biased SNPs and those SNPs with inconsistent quality across DR = 1/ m ∑ ∑( p ij − qij )2 ,
the samples were discarded. This resulted in a total of 511 high j i

quality unbiased SNP loci, which were used for the relationship in which pij and qij are the frequencies of the ith allele at the jth
and structure analysis presented in this paper. locus in each genotype, aj is the number of alleles at the jth locus,
and m is the number of loci examined. Unweighted pair group
method with arithmetic mean (UPGMA)-based phylogeny was
Molecular Statistical Analysis constructed using this Rogers distance matrix. FigTree version
Population substructure was determined using the STRUC-
1.2.3 was used to produce the UPGMA tree image (http://tree.
TURE software (Pritchard et al., 2000). An admixture model
bio.ed.ac.uk/software/figtree [verified 6 Dec. 2010]). Multidi-
with a burn-in time and replication number set at 50,000 was
mensional scaling (MDS) analysis was performed using PROC
used for each run. Three runs were performed for each value
MDS in SAS (SAS Institute, 2003) to evaluate genetic relation-
of K (number of populations) from one to ten. The run with
ships using the Rogers dissimilarity matrix.
the maximum likelihood of the observed genotypes given the
number of subpopulations in the model was used to assign the
probability that a line belongs to each substructure group. Field Evaluations
Summary statistics including total number of alleles, group Trials were grown during the summer of 2008 at the West Madi-
specific alleles, average number of alleles per locus, major allele son Agricultural Research Station in Madison, WI, and sum-
frequency, gene diversity, and polymorphism information con- mer of 2009 at the Arlington Agricultural Research Station
tent (PIC) were calculated using PowerMarker version 3.25 (Liu in Arlington, WI, using a randomized complete block design
and Muse, 2005). Gene diversity (expected heterozygosity) is the with two replications. The 2008 replicated trial contained 611
of the 1411 lines described above. In addition, all 1411 lines were

706 WWW.CROPS.ORG CROP SCIENCE, VOL. 51, MARCH– APRIL 2011

 wet weights were obtained using a John Deere (Deere & Com-
pany, Moline, IL) 5200 self propelled forage harvester converted for
experimental plot harvest. Approximately 500 g of biomass samples
was collected, weighed, and placed in a 55 C° drier for about 7 d.
Samples were then weighed again to calculate percent moisture in
the plot. Whole plot weights were adjusted for percent moisture.
Stalk diameter was measured on the first above-ground elon-
gated internode. Plant height was measured from the soil surface to
the node of the flag leaf. To determine total leaf number a hole was
punched in the fifth leaf of the plants before any leaf senescence. A
collar was then placed between the eighth and ninth leaves before
senescence of the fifth leaf. Internode length was calculated as plant
height divided by leaf number. Ear height was measured from the
soil surface to the node of the upper most developed ear. A devel-
oped ear was defined as one that contained at least five kernels.
Figure 1. Comparison of growing degree day (GDD) accumulation
between 2008 and 2009. Data were collected at the West Madison
Phenotypic Statistical Analysis
Agricultural Research Station (Madison, WI). Planting dates were
A mixed model analysis was performed using PROC GLM in
14 May 2008 and 8 May 2009. SAS (SAS Institute, 2003) to partition variation due to year,
replication, and genotype for each phenotypic trait. For traits
evaluated in nursery plots in 2008. The 2009 replicated trial con- measured on two replications in 2008 and 2009, the model
sisted of the 627 lines in the final Wisconsin diversity set. The included year, replication nested within year, genotype, and
final Wisconsin diversity set was also planted in single replicate in year × genotype interaction. For traits measured on two replica-
2009 at the Madison location, where GDD to flowering time was tions in either 2008 or 2009, the model included replication and
evaluated. All other phenotypic traits were measured on the rep- genotype. For traits measured on one replication in 2008 and
licated trials at Madison and Arlington in 2008 and 2009, respec- 2009, the model included year and genotype. In all models, all
tively. The 2008 Madison replicated experiment was planted on sources of variation were considered random. All statistical tests
14 May. The 2009 Madison single replication experiment was were conducted at a 5% level of significance. PROC CORR
planted on 8 May and the 2009 Arlington replicated experiment was used to determine Pearson correlations across years.
was planted on 23 May. The soil type at both locations is Plano Regression analysis using PROC REG in SAS (SAS Insti-
silt loam (fine-silty, mixed mesic Typic Arguidoll). Temperatures tute, 2003) was conducted to determine the amount of phe-
in 2009 were below average resulting in slower accumulation of notypic variation explained by population substructure. The
GDD (Fig. 1). In 2008, genotypes were evaluated in one-row proportion of membership values for each of the eight popula-
plots (6.1 m long and 0.76 m apart) planted to a density of 79,000 tion substructure groups was included in the regression model.
plants ha−1. In 2009, genotypes at Arlington were evaluated in
two-row plots with the same parameters as in 2008. Genotypes Association Analysis
at Madison in 2009 were evaluated in one-row plots (3.8 m long An insertion or deletion (indel) marker at the Vegetative to generative
and 0.76 m apart) planted at a density of 61,000 plants ha−1. transition 1 (Vgt1) locus that was previously shown to be associated
The following phenotypic traits were measured: GDD to with flowering time in maize (Salvi et al., 2007) was screened
flowering, stover yield on single-row plots (in t ha–1), 300 kernel on 534 of the lines in the diversity panel. The polymerase chain
weight (in g), stalk diameter (in cm), plant height (in cm), leaf reaction (PCR) primers used to amplify the polymorphic regions
number, internode length (in cm), ear height (in cm), upper- were MITE-F (TGGAATGGATGTGAAGTGAG) and MITE-
most internode with a developed ear, number of elongated R (GTACCAGCTCCCTCTGCTC) (Salvi et al., 2007). Poly-
internodes above the uppermost developed ear, last leaf with epi- merase chain reaction products were run on a 1% agarose gel to
cuticular wax, percentage of leaves with epicuticular wax, num- score presence or absence of the mite insertion.
ber of leaves with no epicuticular wax, and percentage of leaves Association mapping was done using the TASSEL soft-
with no epicuticular wax. Stover yield, 300 kernel weight, and ware (Bradbury et al., 2007). A mixed linear model was tested
stalk diameter were only measured in 2008. Uppermost inter- including the eight population substructure covariates and
node with a developed ear and number of elongated internodes appropriate actual proportion of membership in each group (Q
above the uppermost developed ear were only measured in 2009. matrix) generated from STRUCTURE. The kinship matrix
Growing degree days to flowering, stover yield, and 300 kernel was determined using Rogers (1972) distances.
weights were measured on a whole plot basis; all other traits were
measured on five representative plants per plot. RESULTS AND DISCUSSION
Growing degree days to midpollen and midsilk were calcu- Choice of Inbred Lines
lated on a per-plot basis as the average of GDD when half of the
A collection of 1411 inbred lines that would flower in the
plants in the plot were shedding pollen and GDD when half of the
upper Midwest region of the United States was collected.
plants in the plot had exposed silks, respectively. Ears were stripped
from plants at physiological maturity averaged across the experi- This set included 1164 lines from the NC7 as well as 247
ment. Total biomass was harvested from each plot and whole plot additional lines from the Goodman-Buckler diversity set

CROP SCIENCE, VOL. 51, MARCH– APRIL 2011 WWW.CROPS.ORG 707

Table 1. Phenotypic variation of morphological and developmental traits measured on the diversity panel. Values are based on N
inbred lines evaluated in two replications in 2008 (Madison, WI) and in 2009 (Arlington, WI) and one replication in 2009 (Madison, WI).
Genotypes unique to Genotypes in common Complete Wisconsin
Wisconsin diversity set with Goodman-Buckler diversity set diversity set
Phenotypic trait N Min.† Max.† Avg.† R2‡ N Min. Max. Avg. R2 Avg. R2
Days to flowering (growing 383 944.00 1644.90 1277.29 0.18 198 965.09 1644.90 1345.58 0.28 1300.56 0.22
degree days [GDD])§
Stover yield (t ha−1)¶ 338 0.21 4.37 1.69 0.17 190 0.13 3.96 1.99 0.28 1.80 0.20
300 kernel weight (g)§¶ 330 18.50 106.80 71.59 0.18 165 28.39 124.83 67.65 0.10 70.27 0.11
Stalk diameter (cm)¶ 338 1.29 3.10 2.32 0.16 190 1.85 3.25 2.42 0.15 2.36 0.15
Plant height (cm) 398 88.75 239.50 172.23 0.10 206 115.00 232.25 176.76 0.13 173.77 0.09
Leaf number 397 11.92 23.41 18.31 0.14 206 13.90 24.61 19.47 0.28 18.71 0.20
Internode length (cm) 397 5.93 12.82 9.46 0.04 206 5.55 11.75 9.14 0.05 9.35 0.05
Ear height (cm) 398 29.00 144.67 86.84 0.10 206 35.67 143.50 91.69 0.16 88.50 0.12
Uppermost internode with 365 7.50 21.00 13.23 0.14 189 9.20 17.33 13.86 0.21 13.44 0.16
developed ear#
Number elongated inter- 365 1.84 10.50 5.59 0.07 189 3.20 8.10 5.83 0.17 5.67 0.08
nodes above uppermost ear#
Last leaf with epicuticular 398 2.75 13.50 8.01 0.07 206 5.60 11.00 8.21 0.05 8.08 0.03
wax
Percentage of leaves with 397 17.98 70.51 44.14 0.11 206 24.44 64.71 42.70 0.25 43.65 0.13
epicuticular wax
Number leaves with no 397 3.85 17.78 10.31 0.13 206 5.44 17.30 11.27 0.36 10.64 0.20
epicuticular wax
Percentage of leaves with 397 29.49 82.02 55.86 0.11 206 35.29 75.56 57.30 0.25 56.35 0.13
no epicuticular wax
†
Averaged over replications and years.
‡
Phenotypic variation explained by population substructure.
§
Data measured only on one replication per year (all other traits were measured on two replications per year).
¶
Data collected only in 2008.
#
Data collected only in 2009.

and materials provided by the several collaborators. Lines diversity panel contains 15 lines derived from the Germ-
were visually evaluated in a single replication nursery dur- plasm Enhancement of Maize (GEM; Ames, IA) project
ing the summer of 2008 at the West Madison Agricultural (Pollak, 2003). Fourteen of these lines were found to be, by
Research Station in Madison, WI. A subset of 611 of those pedigree, 25% unadapted and 75% elite germplasm and one
lines was selected based on prior performance and pedigree line was 50% unadapted and 50% elite germplasm. When
information and evaluated in a replicated trial at the Madi- assembling a diversity panel, it is important to have a diverse
son location that same year. Based on the 2008 replicated set of germplasm as well as a balance of alleles. Having mul-
evaluation 548 of those lines were further selected to be part tiple genotypes derived from the same open pollinated pop-
of this diversity panel. Selections were based on maturity, ulation helps to maintain a balance of allele frequencies.
agronomic suitability, inbred uniformity, and seed supply.
An additional 79 lines were identified based on the single Phenotypic Diversity
replication nursery evaluation conducted during 2008 and Variation due to genotype was significant for all phenotypic
were added to the set of 548 lines for the 2009 evaluation. traits when the analysis included lines unique to the Wiscon-
The final Wisconsin diversity set therefore contains 627 sin diversity set, lines in common with the Goodman-Buckler
lines (Supplemental Table 1). The entire Wisconsin diversity diversity set, and the complete Wisconsin diversity set (Table
set is available. Plant introduction information is provided 1). Although the subset of the Goodman-Buckler diversity
for the 606 (of the 627) lines available through the NC7. set that was evaluated here is not representative of all pheno-
Source information is provided for the 21 lines not currently typic diversity in maize, it is representative of the phenotypic
available through the NC7 (Supplemental Table 1). diversity that exists in currently characterized inbred lines that
will mature in a short day growing environment. The lines
Relationships by Pedigree unique to this diversity panel expanded the range of pheno-
Many of the Wisconsin diversity set lines trace back to typic variation for many of these traits relative to the subset of
eight open-pollinated populations including Iowa Stiff Stalk lines from the Goodman-Buckler diversity set that met our
Synthetic (BSSS), Minnesota No. 13, Reid Yellow Dent, phenology restriction. For instance, the current panel dramati-
Lancaster Surecrop, Golden Glow, Funk Yellow Dent, cally enhanced the variability for last leaf with epicuticular
Pride of Saline, and Krug among others. In addition, this wax by decreasing the previous minimum by approximately

708 WWW.CROPS.ORG CROP SCIENCE, VOL. 51, MARCH– APRIL 2011

Table 2. Genetic diversity summary statistics based on 511 unbiased single nucleotide polymorphism (SNP) markers. Maturity
groups are based on average flowering time measurements from 2008 and 2009 (Madison, WI) when half of the plants in the
plot were shedding pollen and half of the plants in the plot had exposed silks. Population substructure was determined using
STRUCTURE (Pritchard et al., 2000). Lines in group 9 are mixed genotypes (less than 50% membership within one group).
Total Group Average Major Polymorphism
Set number specific number alleles allele Gene information
size alleles alleles per locus frequency diversity content (PIC)
Category group
Unique to Wisconsin diversity set 411 1008 0 1.9726 0.7857 0.2979 0.2423
Goodman-Buckler diversity set – early 216 1022 0 2.0000 0.7732 0.3163 0.2576
Goodman-Buckler diversity set – late 62 1015 0 1.9863 0.7861 0.3004 0.2448
Goodman-Buckler diversity set 278 1022 14 2.0000 0.7714 0.3197 0.2606
Complete Wisconsin diversity set 627 1022 7 2.0000 0.7796 0.3079 0.2509
Maturity group
1 (less than 75 days after planting) 60 1000 0 1.9569 0.7716 0.3096 0.2496
2 (75–85 days after planting) 201 1005 0 1.9980 0.7822 0.3029 0.2461
3 (85–95 days after planting) 307 1022 0 2.0000 0.7834 0.3036 0.2479
4 (95–105 days after planting) 32 1011 0 1.9785 0.7864 0.3009 0.2455
Population substructure group
1 (SSS† [B73 and B14]) 64 936 0 1.8317 0.8690 0.1792 0.1475
2 (popcorn, sweet corn, and flints) 41 973 0 1.9041 0.7975 0.2772 0.2244
3 (NSS‡ [Wf9]) 8 720 0 1.4090 0.8961 0.1429 0.1157
4 (SSS [B37]) 19 835 0 1.6341 0.8822 0.1677 0.1386
5 (NSS [Minnesota13]) 86 1000 0 1.9569 0.7959 0.2835 0.2301
6 (tropical) 59 1013 0 1.9824 0.7896 0.2939 0.2392
7 (NSS [Oh43 and Iodent]) 39 942 0 1.8434 0.8134 0.2507 0.2020
8 (NSS [Mo17]) 31 933 0 1.8258 0.8481 0.2171 0.1789
9 (mixed) 280 1020 0 1.9961 0.7799 0.3066 0.2494
Wisconsin Diversity set plus Goodman- 689 1022 NA§ 2.0000 0.7775 0.3113 0.2537
Buckler diversity set
†
SSS, Stiff Stalk Synthetic.
‡
NSS, Non-Stiff Stalk Synthetic.
§
NA, not applicable.

three leaves and increasing the previous maximum by 2.5 a similar trend for PIC across maturity groups. Maturity group
leaves. Among other traits, both the upper and lower limit of 3 did not follow these trends, which is likely because there are
variation were expanded for plant height, ear height, upper- significantly more lines in this maturity group. Also, there was
most internode with a developed ear, number of elongated a better balance of alleles in the earlier maturity groups. In the
internodes above the uppermost ear, percentage of leaves with later maturity groups the average major allele frequency across
epicuticular wax, number of leaves with no epicuticular wax, all SNPs is greatest. Maximum power to detect trait loci asso-
and percentage of leaves with no epicuticular wax. The upper ciation occurs when allele frequencies are balanced (Myles et
limit of variation was expanded for stover yield and internode al., 2009). When alleles become rare (<0.05), they are often not
length. The lower limit of variation was expanded for days to included in association mapping or are pooled with other rare
flowering, 300 kernel weight, stalk diameter, and leaf number alleles. In regions where later maturing lines cannot be grown,
(Table 1). A diversity panel intended for association mapping haplotypes from germplasm in the earlier maturity groups are
should utilize the maximum phenotypic diversity possible. the most relevant. Addition of genetic diversity, as indicated by
This increase in phenotypic diversity of characterized inbred gene diversity and PIC based on lines unique to this diversity
lines that will mature in a short day growing environment panel, was greater at early maturity groups and decreased at
for all traits measured except flowering time will increase the later groups due to the consideration of phenology. Although
power to detect trait marker associations for researchers work- the phenology restricted diversity set overall presents a very
ing in short day growing environments. minor reduction in gene diversity (0.3197 to 0.3079) and PIC
(0.2606 to 0.2509) compared to the Goodman-Buckler diver-
Genetic Diversity sity set (Table 2), it will serve as an effective resource for its
Assignment into a maturity group was based on the average intended use.
days after planting (DAP) to flowering from 2008 and 2009. Genetic relatedness was calculated as 1 minus the pair-
The gene diversity was greatest in the earliest maturity group wise Rogers dissimilarity distance (Rogers, 1972) based on
and decreased in the later maturity groups (Table 2). There was 511 unbiased SNP markers (Supplemental Table 2). Only six

CROP SCIENCE, VOL. 51, MARCH– APRIL 2011 WWW.CROPS.ORG 709

Figure 2. Unweighted pair group method with arithmetic mean (UPGMA) tree of 689 diverse inbred lines based on 511 unbiased single
nucleotide polymorphism (SNP) markers. There are 411 genotypes unique to this diversity panel (blue lines), 216 lines from the Goodman-
Buckler diversity set that flower in the upper midwestern United States and are included in this diversity panel (yellow lines), and 62 lines
from the Goodman-Buckler diversity set that do not flower in the upper midwestern United States and are not included in this diversity
panel (red lines). Group labels are based on population substructure analysis. NSS, Non-Stiff Stalk Synthetic; SSS, Stiff Stalk Synthetic.

of the 474,721 pairs had genetic similarity distances less than number of subpopulations in the model had a K value of
0.5 on a scale from 0 to 1. This is likely an artifact of the bial- eight. This run was used to assign the probability that a line
lelic assay used in this analysis. The UPGMA tree constructed belongs to each substructure group based on membership
from these values shows that the current diversity set is repre- probability thresholds of 50%. By pedigree, groups 1 and
sented on all major branches and most subbranches (the blue 4 are primarily Stiff Stalk Synthetic (SSS) lines, group 2 is
and yellow lines in Fig. 2). Several primary branches contain primarily popcorn, sweet corn, and flint lines, groups 3,
only lines unique to this diversity panel (the blue lines in Fig. 2) 5, 7, and 8 are majority Non-Stiff Stalk Synthetic (NSS),
indicating increased representation of many groups. Although and group 6 are tropical and subtropical lines (Fig. 2 and 3,
there is a branch consisting of predominantly red lines denot- Supplemental Table 1, and Supplemental Fig. 1). The group
ing the genotypes from the Goodman-Buckler diversity set 1 SSS lines are mostly B73 and B14 types and the group
that do not meet phenology restriction (the Tropical and Trop- 4 SSS lines are mostly B37 type. Group 3 NSS lines are
ical/Subtropical groups in Fig. 2), the presence of some yellow mostly WF9 type, group 5 NSS lines are mostly Minnesota
lines in this branch indicate that this group is not completely 13 type, group 7 NSS lines are mostly Oh43 and Iodent
excluded in the current panel. Thus, while phenology restric- types, and the group 8 NSS lines are mostly Mo17 type.
tion did not result in complete loss of representation of late The ideal association mapping panel will have subtle pop-
maturity germplasm, it resulted in a major gain in diversity in ulation substructure and familial relatedness (Zhu et al., 2008).
the areas that did not exist previously. False positive associations can be generated in LD analysis due
to the unequal distribution of alleles within subpopulations
Genetic Structure (Flint-Garcia et. al., 2003; Yu et. al., 2006). Thus, it is impor-
The STRUCTURE (Pritchard et al., 2000) run with the tant to determine the genetic substructure and include it in the
maximum likelihood of the observed genotypes given the model. In this diversity panel, population structure accounted

710 WWW.CROPS.ORG CROP SCIENCE, VOL. 51, MARCH– APRIL 2011

Figure 3. Graphical display of population substructure for 627 lines with restricted phenology at population size n = 8. There are 411
genotypes unique to this diversity panel and 216 from the Goodman-Buckler diversity set included in this diversity panel that have
genotypic data. Population substructure was determined using STRUCTURE (Pritchard et al., 2000) with 511 unbiased single nucleotide
polymorphism (SNP) markers. NSS, Non-Stiff Stalk Synthetic; SSS, Stiff Stalk Synthetic.

for between 3% (last leaf with epicuticular wax) and 22% (days group and one within the NSS cloud, consistent with previ-
to flowering [GDD]) of the phenotypic variation (Table 1). ous observations in popcorn lines (Kantety et al., 1995). The
This moderate percentage of phenotypic variation explained NSS lines did not form tight clusters, which is consistent with
by population substructure is desirable. the population substructure results. There are four population
substructures representing the NSS lines while there are only
Comparison between Pedigree two population substructures representing SSS. The MDS plot
and Genetic Relatedness generated from only NSS and SSS classified lines further dem-
Genotypes were grouped into NSS, SSS, tropical, popcorn, onstrates this point (Fig. 5a). When only the NSS and SSS
and sweet corn by pedigree and are color coded on the MDS lines are examined there is very little overlap between the NSS
plot (Fig. 4; Supplemental Table 1). The purpose of MDS plots groups. C103, Mo17, Oh43, Wf9, and PH207 appear in the
is to provide a visual representation of the pattern of related- pedigrees of 58 of the 171 by pedigree NSS lines in this panel.
ness between inbred lines, such that closely related lines will Lines that have one of these founder lines in common by pedi-
be placed near each other on the plot. The SSS, tropical, and gree tend to cluster tightly by genetic distance (Fig. 5a). There
sweet corn classifications form relatively tight clusters on are also subgroups by pedigree that are maintained by genetic
the MDS plot. The popcorn lines included in this diversity distance within the SSS group (Fig. 5b). The SSS subgroups
panel formed two distinct clusters, one cluster near the SSS overlap more than the NSS because B73, B37, and B14 were
all derived from BSSS. The pedigrees of the other SSS do not
trace back to BSSS; however, they do trace back to other Stiff
Stalk Synthetics. For example, there are seven GEM lines that
were crossed to unknown elite inbred lines classified as SSS.
These comparisons indicate that the same population stratifi-
cation observed in maize by pedigree is also observed at the
genetic level.
The relationship between pairs of inbred lines was
determined based on pedigree and genetic analysis. The
parental contribution by pedigree for lines with unknown
origin was considered zero for this analysis. In addition, lines
derived from open pollinated populations were assumed
unrelated due to the lack of sufficient information regarding
the degree of relatedness. This, however, resulted in biased
pair-wise relationships (Supplemental Table 2). To obtain a
correlation value between pedigree and genetic relationships
with minimal bias pair-wise comparisons with a pedigree
Figure 4. Multidimensional scaling plot of 313 lines based on relationship of zero were removed from the analysis. The
511 unbiased single nucleotide polymorphism (SNP) markers. correlation comparison between relatedness by pedigree
Inbred lines with mixed, unrelated, or unknown pedigree were not and relatedness by genotype show a relatively high Pear-
included. Color coded classifications are based on pedigrees. son correlation (R 2 = 0.4892; p < 1 × 10−4) (Fig. 6). There
Coordinates were determined using PROC MDS in SAS (SAS are many reasons that the correlation between genetic and
Institute, 2003) based on a matrix of Rogers distances (Rogers, pedigree distances is significantly less than 1.00. The pedi-
1972). NSS, Non-Stiff Stalk Synthetic; SSS, Stiff Stalk Synthetic. gree distances assume that there is no selection, mutation,

CROP SCIENCE, VOL. 51, MARCH– APRIL 2011 WWW.CROPS.ORG 711

Figure 5. Multidimensional scaling plot based on 511 unbiased single nucleotide polymorphism (SNP) markers. Color coded classifications
are based on pedigrees. Data points with multiple overlapping symbols indicate inbred lines with more than one founder inbred in the
pedigree. Coordinates were determined using PROC MDS in SAS (SAS Institute, 2003) based on a matrix of Rogers distances (Rogers,
1972). Data points with arrow and label indicate the position of the founder line for each subgroup. a) Plot of 266 lines classified by pedigree
as either Stiff Stalk Synthetic (SSS) or Non-Stiff Stalk Synthetic (NSS). Lines were further subdivided into groups based on their pedigree
relatedness to founder lines. b) Plot of 94 lines classified by pedigree as SSS. Lines were classified by pedigree relatedness to founder lines.

712 WWW.CROPS.ORG CROP SCIENCE, VOL. 51, MARCH– APRIL 2011

Figure 6. Correlation between pedigree and genetic relatedness for 2435 pair-wise relationship values. This is a subset of relationship
values from 689 genotypes having 237,016 pair-wise relationship comparisons. Relationships by pedigree with a value of zero were
removed from this plot. There are 411 genotypes unique to this diversity panel, 216 from the Goodman-Buckler diversity set included
in this diversity panel, and 62 lines from the Goodman-Buckler diversity set not included in this diversity panel. Pedigree relatedness
was determined using a modification of the tabular method for calculating coefficients of coancestry that accounts for full inbreeding of
individuals and unequal parental contributions. Genetic relatedness was calculated as pair-wise Rogers distances (Rogers, 1972) based
on 511 unbiased single nucleotide polymorphism (SNP) markers. If lines were derived from sources of unknown origin or open pollinated
populations, the parental contribution from that source was considered zero.

or drift when inbred lines are generated, the first of which insertion was used rather than CGindel587. The mite insertion
is obviously violated in any directed breeding program. In was scored on 534 of the genotypes in this diversity panel.
addition pedigree distances rely solely on identity by descent Flowering time data were collected on one replication in both
and disregard identity by state. Finally, the genetic distances 2008 and 2009. While, genotype × environment interactions
determined by a biallelic assay are likely to be inflated due could not be evaluated, highly significant Pearson correlation
to the lack of allelic states. Although there were ambiguous between the 2008 and 2009 data (r = 0.7576; p < 1 × 10−4)
or unknown pedigrees for some genotypes in this diversity allowed us to use the average data across 2008 and 2009. Asso-
panel this high correlation between genetic and a pedigree ciation analysis using a mixed model that accounts for both
distance suggests the available pedigree information is reli- population substructure and kinship relationships has been
able for many of the lines (Supplemental Table 1). shown to reduce both type I and type II error (Yu et al., 2006).
Using this method, the mite insertion was significantly associ-
Flowering Time Association Analysis ated with flowering time in this population (p = 3.08 × 10−6).
To demonstrate the utility of this diversity panel for None of the 511 SNP analyzed showed any association with
mapping, association analysis was conducted on a region flowering time. This result is expected given the relatively low
known to be linked with flowering time. The flowering marker density in this study. This successful proof of concept
time trait was selected for this proof of concept to demon- related to the mite insertion confirms the utility of this diversity
strate that the phenology restriction placed on the diversity panel for association mapping and lack of any negative conse-
panel did not hinder its utility for association mapping. quences of phenology restrictions. With the added advantage
It has been shown previously that multiple polymor- of inclusion of only those lines maturing in the target area of its
phisms in Vgt1, a noncoding region upstream of ZmRap2.7, use, this panel will be an excellent resource for future associa-
are associated with flowering time in maize (Salvi et al., 2007; tion studies.
Ducrocq et al., 2008). Ducrocq et al. (2008) reported that a
2-bp indel (CGindel587) showed the greatest association with CONCLUSIONS
flowering time in this region. CGindel587 is in high LD with Previously described diversity panels offered limited num-
a mite insertion that is also associated with flowering time ber of genotypes that mature in the upper Midwest region of
(Ducrocq et al., 2008). Due to the ease of genotyping, the mite the United States. Herein, we have described an expanded

CROP SCIENCE, VOL. 51, MARCH– APRIL 2011 WWW.CROPS.ORG 713

set of diverse inbred lines selected based on restricted phe- Breseghello, F., and M.E. Sorrells. 2006. Association mapping of
kernel size and milling quality in wheat (Triticum aestivum L.)
nology and demonstrated that this panel is highly effec- cultivars. Genetics 172:1165–1177.
tive in association mapping studies. Preservation of genetic Camus-Kulandaivelu, L., J. Veyrieras, D. Madur, V. Combes, M.
diversity despite the phenology restrictions coupled with Fourmann, S. Barraud, P. Dubreuil, B. Gouesnard, D. Mani-
the addition of novel genetic and phenotypic variation not cacci, and A. Charcosset. 2006. Maize adaptation to temper-
previously utilized makes this diversity panel a valuable ate climate: Relationship between population structure and
resource to the maize breeding and genetics community. polymorphism in the Dwarf8 gene. Genetics 172:2449–2463.
Casa, A., G. Pressoir, P. Brown, S. Mitchell, W. Rooney, M.
Tuinstra, C. Franks, and S. Kresovich. 2008. Community
Supplemental Information Available
resources and strategies for association mapping in sorghum.
Supplemental material is available free of charge at http:// Crop Sci. 48:30–40.
www.crops.org/publications/cs. Crossa, J., J. Burgueño, S. Dreisigacker, M. Vargas, S. Herrera-
Foessel, M. Lillemo, R. Singh, R. Trethowan, M. Warbur-
Acknowledgments ton, J. Franco, M. Reynolds, J. Crouch, and R. Ortiz. 2007.
The authors would like to acknowledge Dustin Eilert, Julianne Association analysis of historical bread wheat germplasm
Smith, Bill Kojis, and Bob Vogelzang for technical assistance, using additive genetic covariance of relatives and population
Ed Buckler and Jeff Glaubitz for their help in designing the structure. Genetics 177:1889–1913.
Illumina GoldenGate high-throughput assay, Rex Bernardo Ducrocq, S., C. Giauff ret, D. Madur, V. Combes, F. Dumas, S.
for his help with the RELATE program, and Sherry Flint- Jouanne, D. Coubriche, P. Jamin, L. Moreau, and A. Char-
Garcia for thoughtful suggestions regarding this manuscript. cosset. 2009. Fine mapping and haplotype structure analysis
We are also grateful for support provided by the D.C. Smith of a major flowering time quantitative trait locus on maize
and Gabelman-Shippo Graduate Fellowship programs at the chromosome 10. Genetics 183:1555–1563.
University of Wisconsin – Madison. This work was supported Ducrocq, S., D. Madur, J. Veyrieras, L. Camus-Kulandaivelu, M.
by the U.S. Department of Energy through the Great Lakes Kloiber-Maitz, T. Presterl, M. Ouzunova, D. Manicacci, and
Bioenergy Research Center Grant DE-FC02-07ER64494 and A. Charcosset. 2008. Key impact of Vgt1 on flowering time
USDA-Hatch projects WIS01330 and WIS01335. adaptation in maize: Evidence from association mapping and
ecogeographical information. Genetics 178:2433–2437.
Ehrenreich, I., P. Stafford, and M. Purugganan. 2007. The genetic
References architecture of shoot branching in Arabidopsis thaliana: A com-
Agrama, H., G. Eizenga, and W. Yan. 2007. Association mapping of parative assessment of candidate gene associations vs. quanti-
yield and its components in rice cultivars. Mol. Breed. 19:341–356. tative trait locus mapping. Genetics 176:1223–1236.
Andersen, J.R., T. Schrag, A. Melchinger, I. Zein, and T. Lübber- Emik, L.O., and C.E. Terrill. 1949. Systematic procedures for cal-
stedt. 2005. Validation of Dwarf8 polymorphisms associated culating inbreeding coefficients. J. Hered. 40:51–55.
with flowering time in elite European inbred lines of maize Fan, J., M. Chee, and K. Gunderson. 2006. Highly parallel
(Zea mays L.). Theor. Appl. Genet. 111:206–217. genomic assays. Nat. Rev. Genet. 7:632–644.
Aranzana, M., S. Kim, K. Zhao, E. Bakker, M. Horton, K. Jakob, Flint-Garcia, S., J. Thornsberry, and E. Buckler. 2003. Structure
C. Lister, J. Molitor, C. Shindo, C. Tang, C. Toomajian, B. of linkage disequilibrium in plants. Annu. Rev. Plant Biol.
Traw, H. Zheng, J. Bergelson, C. Dean, P. Marjoram, and 54:357–374.
M. Nordborg. 2005. Genome-wide association mapping in Flint-Garcia, S., A. Thuillet, J. Yu, G. Pressoir, S. Romero, S. Mitch-
Arabidopsis identifies previously known flowering time and ell, J. Doebley, S. Kresovich, M. Goodman, and E. Buckler.
pathogen resistance genes. PLoS Genet. 1:E60. 2005. Maize association population: A high-resolution platform
Bao, J., H. Corke, and M. Sun. 2006. Microsatellites, single for quantitative trait locus dissection. Plant J. 44:1054–1064.
nucleotide polymorphisms and a sequence tagged site in Gerdes, J., C. Behr, J. Coors, and W. Tracy. 1993. Compilation of North
starch-synthesizing genes in relation to starch physicochemi- American maize breeding germplasm. CSSA, Madison, WI.
cal properties in nonwaxy rice (Oryza sativa L.). Theor. Appl. Harjes, C., T. Rocheford, L. Bai, T. Brutnell, C. Kandianis, S.
Genet. 113:1185–1196. Sowinski, A. Stapleton, R. Vallabhaneni, M. Williams, E.
Beló, A., P. Zheng, S. Luck, B. Shen, D. Meyer, B. Li, S. Tingey, Wurtzel, J. Yan, and E. Buckler. 2008. Natural genetic varia-
and A. Rafalski. 2008. Whole genome scan detects an allelic tion in Lycopene Epsilon Cyclase tapped for maize biofortifica-
variant of fad2 associated with increased oleic acid levels in tion. Science 319:330–333.
maize. Mol. Genet. Genomics 279:1–10. Jun, T., K. Van, M. Kim, S. Lee, and D. Walker. 2008. Associa-
Bernardo, R., A. Murigneux, J. Maisonneuve, C. Johnsson, and tion analysis using SSR markers to fi nd QTL for seed protein
Z. Karaman. 1997. RFLP-based estimates of parental contri- content in soybean. Euphytica 162:179–191.
bution to F2 - and BC1-derived maize inbreds. Theor. Appl. Kantety, R., X. Zeng, J. Bennetzen, and B. Zehr. 1995. Assess-
Genet. 94:652–656. ment of genetic diversity in dent and popcorn (Zea mays L.)
Botstein, D., R. White, M. Skolnick, and R. Davis. 1980. Construc- inbred lines using inter-simple sequence repeat (ISSR) ampli-
tion of a genetic linkage map in man using restriction fragment fication. Mol. Breed. 1:365–373.
length polymorphisms. Am. J. Hum. Genet. 32:314–331. Kraakman, A., F. Martínez, B. Mussiraliev, F. van Eeuwijk, and
Bradbury, P., Z. Zhang, D. Kroon, T. Casstevens, Y. Ramdoss, and R. Niks. 2006. Linkage disequilibrium mapping of morpho-
E. Buckler. 2007. TASSEL: Software for association mapping of logical, resistance, and other agronomically relevant traits in
complex traits in diverse samples. Bioinformatics 23:2633–2635. modern spring barley cultivars. Mol. Breed. 17:41–58.
Liu, K., M. Goodman, S. Muse, J. Smith, E. Buckler, and J. Doebley.

714 WWW.CROPS.ORG CROP SCIENCE, VOL. 51, MARCH– APRIL 2011

 2003. Genetic structure and diversity among maize inbred lines Stracke, S., G. Haseneyer, J. Veyrieras, H. Geiger, S. Sauer, A.
as inferred from DNA microsatellites. Genetics 165:2117–2128. Graner, and H. Piepho. 2009. Association mapping reveals
Liu, K., and S.V. Muse. 2005. PowerMarker: Integrated analysis envi- gene action and interactions in the determination of flower-
ronment for genetic marker data. Bioinformatics 21:2128–2129. ing time in barley. Theor. Appl. Genet. 118:259–273.
McMullen, M., S. Kresovich, H. Villeda, P. Bradbury, H. Li, Q. Szalma, S., E. Buckler, M. Snook, and M. McMullen. 2005. Associa-
Sun, S. Flint-Garcia, J. Thornsberry, C. Acharya, C. Bot- tion analysis of candidate genes for maysin and chlorogenic acid
toms, P. Brown, C. Browne, M. Eller, K. Guill, C. Harjes, accumulation in maize silks. Theor. Appl. Genet. 110:1324–1333.
D. Kroon, N. Lepak, S. Mitchell, B. Peterson, G. Pressoir, S. Thornsberry, J., M. Goodman, J. Doebley, S. Kresovich, D. Nielsen,
Romero, M. Rosas, S. Salvo, H. Yates, M. Hanson, E. Jones, and E. Buckler. 2001. Dwarf8 polymorphisms associate with
S. Smith, J. Glaubitz, M. Goodman, D. Ware, J. Holland, and variation in flowering time. Nat. Genet. 28:286–289.
E. Buckler. 2009. Genetic properties of the maize nested asso- Wang, J., P. McClean, R. Lee, J. Goos, and T. Helms. 2008a.
ciation mapping population. Science 325:737–740. Association mapping of iron deficiency chlorosis loci in soy-
Mikel, M.A. 2006. Availability and analysis of proprietary dent bean (Glycine max L. Merr.) advanced breeding lines. Theor.
corn inbred lines with expired U.S. plant variety protection. Appl. Genet. 116:777–787.
Crop Sci. 46:2555–2560. Wang, R., Y. Yu, J. Zhao, Y. Shi, Y. Song, T. Wang, and Y. Li.
Mikel, M.A., and J.W. Dudley. 2006. Evolution of North Ameri- 2008b. Population structure and linkage disequilibrium of a
can dent corn from public to proprietary germplasm. Crop mini core set of maize inbred lines in China. Theor. Appl.
Sci. 46:1193–1205. Genet. 117:1141–1153.
Murray, S., W. Rooney, M. Hamblin, S. Mitchell, and S. Kreso- Warburton, M., X. Xianchun, J. Crossa, J. Franco, A. Melchinger,
vich. 2009. Sweet sorghum genetic diversity and association M. Frisch, M. Bohn, and D. Hoisington. 2002. Genetic char-
mapping for brix and height. Plant Gen. 2:48–62. acterization of CIMMYT inbred maize lines and open pol-
Myles, S., J. Peiffer, P. Brown, E. Ersoz, Z. Zhang, D. Costich, linated populations using large scale fi ngerprinting methods.
and E. Buckler. 2009. Association mapping: Critical consid- Crop Sci. 42:1832–1840.
erations shift from genotyping to experimental design. Plant Wei, X., P. Jackson, C. McIntyre, K. Aitken, and B. Croft. 2006.
Cell 21:2194–2202. Association between DNA markers and resistance to diseases
Olsen, K., S. Halldorsdottir, J. Stinchcombe, C. Weinig, J. Schmitt, in sugarcane and effects of population substructure. Theor.
and M. Purugganan. 2004. Linkage disequilibrium mapping of Appl. Genet. 114:155–164.
Arabidopsis CRY2 flowering time alleles. Genetics 167:1361–1369. Weir, B.S. 1996. Genetic data analysis II. Sinauer Associates, Sun-
Olsen, K.M., and M.D. Purugganan. 2002. Molecular evidence on derland, MA.
the origin and evolution of glutinous rice. Genetics 162:941–950. Wilson, L., S. Whitt, A. Ibáñez, T. Rocheford, M. Goodman, and
Pollak, L.M. 2003. The History and success of the public–private E. Buckler. 2004. Dissection of maize kernel composition and
project on germplasm enhancement of maize (GEM). Adv. starch production by candidate gene association. Plant Cell
Agron. 78:45–87. 16:2719–2733.
Pressoir, G., P. Brown, W. Zhu, N. Upadyayula, T. Rocheford, E. Xie, C., M. Warburton, M. Li, X. Li, M. Xiao, Z. Hao, Q. Zhao,
Buckler, and S. Kresovich. 2009. Natural variation in maize and S. Zhang. 2008. An analysis of population structure and
architecture is mediated by allelic differences at the PINOID linkage disequilibrium using multilocus data in 187 maize
co-ortholog barren inflorescnece2. Plant J. 58:618–628. inbred lines. Mol. Breed. 21:407–418.
Pritchard, J., M. Stephens, and P. Donnelly. 2000. Inference of Yamasaki, M., S. Wright, and M. McMullen. 2007. Genomic
population structure using multilocus genotype data. Genet- screening for artificial selection during domestication and
ics 155:945–959. improvement in maize. Ann. Bot. (London) 100:967–973.
Purdy, L., W. Loegering, C. Konzak, C. Perterson, and R. Allan. Yan, J., T. Shah, M. Warburton, E. Buckler, M. McMullen, and J.
1968. A proposed standard method for illustrating pedigrees Crouch. 2009. Genetic characterization and linkage disequi-
of small grain varieties. Crop Sci. 8:405–406. librium estimation of a global maize collection using SNP
Reif, J., X. Xia, A. Melchinger, M. Warburton, D. Hoisington, markers. PLoS One 4:E8451.
D. Beck, M. Bohn, and M. Frisch. 2004. Genetic diversity Yu, J., G. Pressoir, W. Briggs, I. Bi, M. Yamasaki, J. Doebley, M.
determined within and among CIMMYT maize populations McMullen, B. Gaut, D. Nielsen, J. Holland, S. Kresovich, and
of tropical, subtropical, and temperate germplasm by SSR E. Buckler. 2006. A unified mixed-model method for associa-
markers. Crop Sci. 44:326–334. tion mapping that accounts for multiple levels of relatedness.
Rogers, J.S. 1972. Measure of genetic similarity and genetic dis- Nat. Genet. 38:203–208.
tance. Studies in Genetics. VII. Univ. Tex. Publ. 7213:145–153. Zhao, K., M. Aranzana, S. Kim, C. Lister, C. Shindo, C. Tang, C.
Saghai-Maroof, M., K. Soliman, R. Jorgensen, and R. Allard. Toomajian, H. Zheng, C. Dean, P. Marjoram, and M. Nord-
1984. Ribosomal DNA spacer-length polymorphisms in bar- borg. 2007. An Arabidopsis example of association mapping in
ley: Mendelian inheritance, chromosomal location, and pop- structured samples. PLoS Genet. 3:E4.
ulation dynamics. Proc. Natl. Acad. Sci. USA 81:8014–8018. Zheng, P., W.B. Allen, K. Roesler, M.E. Williams, S. Zhang, J.
Salvi, S., G. Sponza, M. Morgante, D. Tomes, X. Niu, K. Fengler, Li, K. Glassman, J. Ranch, D. Nubel, W. Solawetz, D. Bhat-
R. Meeley, E. Ananiev, S. Svitashev, E. Bruggemann, B. Li, C. tramakki, V. Llaca, S. Deschamps, G.-Y. Zhong, M.C. Tarc-
Hainey, S. Radovic, G. Zaina, J. Rafalski, S. Tingey, G. Miao, R. zynski, and B. Shen. 2008. A phenylalanine in DGAT is a key
Phillips, and R. Tuberosa. 2007. Conserved noncoding genomic determinant of oil content and composition in maize. Nat.
sequences associated with a flowering-time quantitative trait Genet. 40:367–372.
locus in maize. Proc. Natl. Acad. Sci. USA 104:11376–11381. Zhu, C., M. Gore, E. Buckler, and J. Yu. 2008. Status and pros-
SAS Institute. 2003. The SAS system for windows version 9.1. SAS pects of association mapping in plants. Plant Gen. 1:5–20.
Inst, Cary, NC.

CROP SCIENCE, VOL. 51, MARCH– APRIL 2011 WWW.CROPS.ORG 715