Download Molecular population genetics and agronomic

Journal of Experimental Botany, Vol. 60, No. 9, pp. 2541–2552, 2009
doi:10.1093/jxb/erp130 Advance Access publication 18 May, 2009
REVIEW PAPER
Molecular population genetics and agronomic alleles in seed
banks: searching for a needle in a haystack?
Dario Prada*
Seed Conservation Department, Royal Botanic Gardens Kew, Wakehurst Place, West Sussex RH17 6TN, UK
Received 14 January 2009; Revised 31 March 2009; Accepted 31 March 2009
Abstract
Seed banking has been the single most significant reaction of the research community to the alarming rates of plant
genetic erosion occurring in the wild. One enduring challenge for a wiser utilization of the resources enclosed in
seed banks, however, has been the estimation of their genetic potentials for agriculture’s benefit. Key to detecting in
landraces and/or wild relatives of modern crops any allelic variant lost during domestication and crop improvement
is the use of molecular information to determine structure, evolution, and function of the genes harbouring these
alleles. This paper reviews some of the theoretical and statistical issues surrounding the use of molecular population
genetics tools for the detection of agronomical valuable alleles in seed banks. Emphasis is made on the technical
limitations imposed by seed banking that may lessen the success of integrated and multi-disciplinary molecular
approaches. The influence that population stratification and linkage disequilibrium exert on specific experimental
designs for a better understanding of the evolutionary history of potential agronomic-related genes is also examined.
Key words: Association mapping, linkage disequilibrium, plant genetic resources, plant molecular genetics, population
stratification, seed banking, selective genes.
Seed banking and its impact on modern agriculture
Plant improvement relies on selective propagation of
genotypes containing favourable combinations of alleles of
genes controlling desirable agronomic traits. Over the last
10 000 years plant domestication produced numerous
landrace populations that served as the founder material
for further genetic improvement through more recent
selective breeding in the last 150 years. Both early domestication and later crop improvement have caused several
genetic bottlenecks presumably reducing the levels of
genetic diversity in modern crops. In fact, most of the
contemporary crop varieties descend from a relatively small
number of founder landraces. It follows that those genes
underlying agronomic traits in modern cultivated genotypes
retain decreased levels of diversity compared with the entire
gene pool of landrace populations and the closest wild
relatives from which they derive (Tanksley and McCouch,
1997). For example, in the case of grasses, the world’s most
important crop family, cultivated varieties maintain, on
average, 70% of the diversity found in their closest wild
progenitors (Buckler et al., 2001).
Nowadays there is a long-standing concern in the international community about the disappearance of genetic
variants (Esquinas-Alcazar, 2005). The historic narrowing
of the genetic basis for enhanced agronomic performance
might make, for instance, modern varieties more susceptible
to newly emerging diseases (Harlan, 1975). Global climate
change over the last ;30 years has produced directional
shifts in the distribution and abundance of wild plant
communities, representing a major cause of widespread
reduction of the biological diversity (Parmesan and Yohe,
2003). In reaction to these threats of genetic erosion, the
global strategy adopted by the research community has
been to increase efforts to warehouse wild plant species in
seed banks (Schoen and Brown, 2001). Seed banking
prolongs seed viability ex situ under cold and dry conditions
and thus safeguards plants for future use. The collections
* Present address: Pioneer Hi-Bred International, Via Madre Teresa di Calcutta n.2/4, Pessina Cremonese, CR 26030, Italy. E-mail: [email protected]
ª The Author [2009]. Published by Oxford University Press [on behalf of the Society for Experimental Biology]. All rights reserved.
For Permissions, please e-mail: [email protected]
2542 | Prada
held in seed banks are usually formed by gene pools of
individual crops at the interspecies level, that is, cultivated
crops, their founder ancestors as well as the closest-relative
species that still survive in the wild. In 1997, for instance,
FAO reported that the seed bank collections maintained by
the Consultative Group on International Agricultural Research (CGIAR) consisted of 59% landraces and old
cultivars, 27% modern and historical breeders’ cultivars,
and 14% wild relatives (FAO, 1997). In the case of major
crops held in seed banks, there are many examples of wild
relatives organized in the form of regional accessions, such
as wild barley (Hordeum spontaneum), wild maize (five
recognized species of teosinte for the genus Zea), wild
wheat (Triticum urartu, T. boeoticum, and T. dicoccoides),
wild rice (Oryza barthii, O. glumaepatula, O. meridionalis,
O. nivara, and O. rufipogon), or wild rye (Secale vavilovii).
There can also be found in seed banks some collections
derived from crosses between taxonomically-related crops at
the interspecies level; one of the most significant examples
of this type of germplasm is the ‘Veery wheat’ lines which
were obtained after the introgression of an entire chromosomal segment of rye into wheat (Rajaram et al., 1983;
Hoisington et al., 1999). As a whole, the seed bank
collections may quantitatively represent only a small fraction
of the global biodiversity, but they are undoubtedly one of
the world’s richest stocks of plant genetic diversity and offer
a source of alleles for future genetic improvement of crops.
Despite the preponderant role of seed banks in sheltering
novel alleles with potential use for crop enhancement, the
reality is that seed bank accessions still have a very limited
impact on current crops (FAO, 1997). Some technical
restrictions associated with seed banking may have lessened
the use of wild plant genetic resources into agriculture.
First, since seed bank collections are stored outside the
natural habitat, typically in such a way as to minimize
genetic change, they show a lack of evolutionary response
to changing environmental conditions with its resulting low
adaptation rate. Second, because of practical (small)
collection sizes, founder effects may possibly induce changes
in frequency distribution of the alleles present in seed bank
accessions—samples are usually obtained from fewer than
100 individuals in the wild. Third, seed viability declines
during ex situ storage, hence the collections require regeneration in order to replenish stocks; however, regeneration strategies too often contribute to a shift in the genetic
composition (genetic drift) of the accessions. Fourth, the
ability to implement and design strategies for the identification and isolation of useful novel genes in wild donors has
been proven unsuccessful in many ways. In this context, the
development of efficient strategies that can facilitate the
active incorporation of wild genetic resources into agricultural systems still remains an active area of research.
Strategies for the isolation of agronomical
desired genes in seed banks
Traditionally, the approach for the utilization of the genetic
material maintained in seed banks has been to screen
accessions for a phenotypic appearance. Individual wild
progenitors or landraces having the desired phenotype are
repeatedly backcrossed with an elite genotype eventually to
introduce the beneficial wild allele into the cultivated genetic
background. Remarkable examples of recurrent backcross
have been primarily produced for Lycopersicon. Within this
genus, some wild species can be effectively crossed with the
cultivated tomato L. esculentum and have been successfully
used as donors of fungus- and insect-resistant genes (L.
hirsutum and L. peruvianum), genes for fruit quality
improvement (L. chmielewskii), and genes for adaptation to
adverse environments (L. cheesmaniae) (Esquinas-Alcazar,
2005). Crosses between cultivated forms of sugar beet and
its wild relative species have also been made to enhance
disease-resistance and broaden the genetic base of the crop
in general (Fig. 1; sugar beet). The repeated backcross
strategy looks first at phenotypic observations to characterize the underlying genetic architecture: it starts the analysis
at high levels of biological organization (phenotype) and
descends to lower levels in order to gain knowledge on how
genetic variation is arrayed within crops and relatives (the
so-called ‘top-down’ approach). However, this approach
has serious limitations because it is only applicable to genes
easily observable in the phenotype which are mostly
controlled by one, or a few, set(s) of genes.
In contrast to ‘top-down’ genetics, alternative approaches
that begin the analyses at the genomic sequence level and
then work back up to the phenotype (so-called ‘bottom-up’
strategies) have emerged recently with real promise for
a more comprehensive understanding of plant genetic
variation. Tanksley and McCouch (1997), in a seminal
paper, stressed the poor prediction capacity of the phenotype and pointed out the genetic composition at molecular
level as the best indicator of genetic potentials in seed bank
accessions. The approach they suggested centred on linkage
mapping analysis over populations derived from crosses
between a crop variety and one of its wild relatives: once
a target quantitative trait locus (QTL) is identified in the
mapping population, advanced molecular marker-assisted
backcrossing facilitates the isolation of the causal wild
beneficial allele within a homogeneous elite genetic background (Fig. 1; tomato). This method may theoretically
allow more accurate resolution of any specific QTL, in
principle, from typical 10 cM QTL intervals into 1 cM
intervals, identifying targets for further positional cloning.
Nevertheless, the progress in isolating, cloning, and characterizing novel genes following this strategy has been restricted to a small number of noteworthy cases, such as tb1
in maize (Doebley et al., 1995), fw2.2 in tomato (Frary
et al., 2000), and Hd1 in rice (Yano et al., 2000). What has
prevented this pioneering approach from fulfilling its initial
expectations? The lack of more positive results could be
attributed to different facts. On the one hand, the modest
degree of recombination in practical population sizes may
have limited the power of the statistical tests for QTL
detection and, on the other hand, since only two alleles are
sampled in a biparental population, there is a clear underrepresentation of the putative pool of allelic variants at
Plant molecular population genetics and seed banks | 2543
Fig. 1. Historical strategies for the deployment and isolation of potentially valuable agronomic alleles enclosed in the genetic background
of wild relatives of crops.
a locus. Moreover, even if statistically significant evidence
of linkage is obtained, extensive positional cloning should
still be required to progress from a broad linkage region
containing millions of bases to the causal gene(s) within the
genomic region.
In the hope of overcoming some of the limitations of
linkage analysis, the plant research community has only
recently started to exploit the natural genetic diversity of
germplasm collections as an additional means to identify
marker-trait associations. This type of (association) analysis
is performed across highly diverse sets of genotypes which
contain many more historical recombinational events than
the biparental populations used in linkage analysis, allowing
much higher mapping resolution. The target germplasm
collections for association tests often include elite and
historic commercial cultivars, as well as landraces and the
wild relatives of crops (Zhu et al., 2008). Different
accessions of teosinte (Zea mays ssp. parviglumis), the wild
progenitor of maize, have been used, for instance, in
association mapping panels for the study of major regulatory genes controlling plant growth and development
(Fig. 1; maize). One reason for optimism regarding
‘bottom-up’ techniques (i.e. linkage and association mapping) in disclosing wild genetic potentials is that they can
play a critical role in moving from descriptive observations
in the phenotype, extensively used until now to predictive
functional diversity on the basis of polymorphisms at DNA
level.
Molecular population genetics and plant
genetic potentials in seed banks
Plant researchers are beginning to benefit from some of the
specific areas of research falling under the molecular
population genetics framework, with evolutionary biology,
association mapping, and comparative genomics being three
of its most representative examples. These methods embody
‘bottom-up’ approaches used in agricultural research (also
called ‘reverse’ genetics). In the sections that follow, some of
the theoretical, statistical, and practical issues surrounding
the efficient deployment of molecular population genetics
tools for the detection of agronomically valuable alleles in
seed banks are presented. The stepwise integration of these
techniques may signal the advent of a new and promising
era for a better understanding and use of the genetic
variation enclosed in seed banks. Figure 2 summarizes
possible interconnections among different molecular genetics applications in order to gain better genetic knowledge of
the novel and (valuable) functional variation present in seed
banks. This general methodology, based on combined tests
of selection and association mapping, is being used in
humans for the isolation and characterization of specific
genomic sequences; in plant research it has also been
applied but with various levels of integration and
resolution.
Cultivated genotypes are the direct result of the accumulation of beneficial alleles at key genes controlling traits of
agronomic interest, but considering the high number of loci
expected to participate in the phenotypic expression of such
traits, it is very unlikely that modern genotypes retain
beneficial allelic variants at all of the agronomically related
sites (Tanksley and McCouch, 1997). It is thought that sets
of interesting classes of alleles, in many cases with minor
effects, could have been missed during domestication and/or
crop improvement. The evolutionary history of agronomically related genes has been shaped primarily by humanmediated selection events, either unintentionally through
domestication or intentionally through more recent crop
improvement activities. The resulting genetic bottlenecks
2544 | Prada
Fig. 2. Scheme using a modified Forrester’s symbol system (Forrester, 1961) of combined molecular population genetics methods for
sequence identification, functional validation, and further introgression in modern crops of novel alleles present in wild relatives and
landraces at seed banks. Sources of material are written in red (WR, wild relative; LR, landrace; MC, modern crop). Ellipses are specific
approaches, valves are outcomes for decision, and arrows are flows of information. A, B, C, and D are referred to in the text.
created in this course have very likely altered the allelic
frequencies at loci with agronomical significance, as well as
inducing overall genetic variation to decrease and linkage
disequilibrium (LD) to increase (Fig. 2A). LD refers to the
non-zero correlation between alleles at different loci, even
unlinked, and it is inversely proportional to levels of allelic
recombination (Flint-Garcia et al., 2003). The extent of LD
is expected to vary within individual genetic pools forming
the seed bank accessions; LD seems to decrease gradually
for modern genotypes, landraces, and wild relatives. Such
differences are mainly explained by the diverse mating
histories of each genetic pool. The decay of LD across
different germplasm pools has been extensively characterized for major annual crops (Gupta et al., 2005). For
example, for modern cultivated varieties of barley, complete
LD has been observed across contiguous genomic sequences
of up to 212 kb in length (Piffanelli et al., 2004) while in
landraces LD decays over 90 kb and in wild barley LD does
not extend beyond a single genic region (Caldwell et al.,
2006). Analogous patterns of LD decay have also been
observed in maize: it extends 100 kb for commercial elite
inbred lines but declines to levels of 1.5 kb for maize
landraces (Yu and Buckler, 2006). There are only a few
studies that have estimated the extent of LD in perennials,
in part, because these species have more complex life
histories and mating systems than annual crops. Examples
can be found for woody plants—European aspen (Populus
tremula) (Ingvarsson, 2005), Scots pine (Pinus sylvestris)
(Dvornyk et al., 2002), Douglas fir (Pseudostuga menziensii)
(Krutovsky and Neale, 2005), and loblolly pine (Pinus
taeda) (Brown et al., 2004)—as well as for grapevine (Vitis
vinifera) (Barnaud et al., 2006) and perennial ryegrass
(Lolium perenne) (Ponting et al., 2007).
The onset of ‘agriculture’ occurred in the wake of
unconscious selection made by humans, leading to the
fixation of relevant alleles in a reduced set of key traits with
drastic impact on the phenotype, such as seed size, ear
rachis stiffness, and the ease with which the seed was
released from its enclosing leaf-like structures. Afterwards,
there was a long phase of human-mediated selection over
primitive wild plant forms that allowed the expansion of
agriculture into new environments, in this case mainly by
modifying different, but again relatively small, sets of traits
with polygenic inheritance (i.e. seed weight, seed dormancy)
which resulted in landrace populations adapted to local
conditions (Salamini et al., 2005). Common patterns of
selection seem to have been applied to a limited group of
traits/genes throughout more recent breeding activities.
Interestingly, for the most important crop families, target
loci for plant improvement show high levels of homology
across species. For example, in the case of cereals, the
homologue sequences of the genes RhtD1a of wheat and
Plant molecular population genetics and seed banks | 2545
Gai of rice, both affecting plant height and flowering time,
had impressive effects on yield during the ‘Green Revolution’ in the 1960s and 1970s (Peng et al., 1999). Flowering
time has had a central role in selection for local adaptation
across several major crops; landrace populations of cereals
that flowered in short days have been transformed during
selection into crops in which flowering time is unaffected by
day length (Putterill et al., 2004). This is the case for barley,
in which the gene Ppd-H1 that controls flowering time had
a strong influence in the domestication and further Neolithic spread of cultivation due to human-mediated selection
of non-responsive ppd-H1 phenotypes (Jones et al., 2008).
This convergent selection in cereals may have been centred
on a set of genes homologous to the PHY family of
Arabidopsis thaliana (Sawers et al., 2005).
Modern crops constitute good models to evaluate
imprints of human-mediated selection at specific genomic
sites, frequently sharing homology across species, especially
for the traits where domestication and/or breeding have
acted more strongly (Wright and Gaut, 2005). The reduction in allelic diversity and enhancement of LD have
been extensively used in plant molecular population genetics
as primary signals of selection on random sequence data
(Fig. 2B). Wright et al. (2005) suggest that if genomic sites
showing clear evidence of selection are identified in modern
crops they may provide a substrate for the amplification of
homologous alleles in sets of landraces and wild relatives.
These sequences ultimately represent allelic forms missed in
modern crops because of selection at different time scales,
but that still exist in the wild as candidates for novel
variation in agronomically related genes. Primer design,
however, can be very difficult for the very heterogeneous
germplasm kept in seed banks because priming in more
conserved regions may amplify paralogous regions while
priming in less conserved regions may fail if there is
extensive polymorphism in the primer region across the
germplasm (Flint-Garcia et al., 2005). In spite of these wellknown technical drawbacks, amplification of homologous
sequences across different genetic pools at interspecies level
has proved to be successful on the maize genome. Common
primer sequences that are maintained for modern corn
varieties with respect to teosinte, its wild common ancestor,
have been amplified in the wild ancestor background and
have been subsequently flagged for introgression into the
elite germplasm pool of maize (Wright et al., 2005).
The influence of selective genes on phenotype varies
tremendously. As a result, not all selective alleles should
necessarily have a real impact over agronomical traits. In
order to avoid novel but non-functional (non-valuable)
genetic variation being reintroduced into modern crops, the
effect of candidate loci on the phenotype needs to be
investigated. At present, association analysis is the most
powerful method in molecular population genetics to
establish the link between genotype and phenotype in highly
diverse panels (Fig. 2C). If candidate sequences for introgression are eventually identified and validated, any
molecular marker-assisted backcross strategy can then
facilitate the reintroduction process by monitoring the
presence of target primer sequences as backcross generations advance (Fig. 2D). Crosses between cultivated genotypes and their wild relatives within the same genus, thus
overcoming the sexual barriers for inter-specific crossing,
have been consistently produced for several major crops like
tomato, sugar beet, maize, and barley. Molecular markerassisted backcross schemes are routinely used in modern
agricultural research; a classical example is Marker-Assisted
Recurrent Selection (MARS) which refers to the improvement of an F2 population by combining several cycles of
phenotypic and marker selection (Johnson, 2004).
All molecular genetics techniques described in Fig. 2 have
been effectively applied in plant research but with different
levels of integration. Maize and tomato are the crops on
which most research has been devoted in recent years. There
are two central aspects yet to consider when linking up any
molecular genetics technique through multi-disciplinary approaches, the existence of stratification due to population
admixture of the sample and the degree of LD (populationspecific) between alleles. In the next sections, the implications
that these two phenomena may have in the detection of signals of selection, and in functional validation, are examined.
Identification of alleles with imprints of
selection in seed banks
Two types of selection have been targeted in crop evolution,
i.e. either directional or balancing selection. Directional
selected genes are normally genes with high influence on the
phenotype and, in many cases, have been easily selected by
obvious morphological and developmental phenotypic
observations. This is the case for the tb1 gene in maize that
governs lateral branching (Doebley et al., 1997) and the
RhtD1a gene in wheat that controls plant height and
flowering time (Peng et al., 1999). For direct deployment in
breeding programmes, directional-selected genes that have
undergone the most stringent selection (and thereby the
greatest reduction in diversity) have little remaining genetic
variation and cannot easily be further improved by breeding. In contrast, genes bearing signatures of balancing
selection may be of greater importance because valuable
allelic variants having small impact on the phenotype could
still reside in the wild. Directional selection, also referred to
as selective sweep, favours one allele over others and can
lead to the fixation of the favoured variant in the entire
population. Balancing selection implies the long-term selective maintenance of multiple allelic variants at intermediate
frequencies not resulting in allele fixation. The simplest
model to explain most evolutionary change in the wild
through selection was proposed by Kimura (1968). His
neutral equilibrium theory (NE) suggests that plant adaptation is the result of genetic drift acting on selectivelyequivalent (neutral) mutant alleles. NE serves then as the
null hypothesis to evaluate imprints of selection through
modification of allelic frequencies from equilibrium expectations. Following Kimura’s work, many statistical tests
have been developed to screen sequence data for differences
2546 | Prada
in allelic distributions relative to NE expectations. The most
popular statistical indicator is the mean value of Tajima’s D
statistic (Tajima, 1989) which compares the number of
nucleotide polymorphisms with the mean pairwise difference between sequences. Other alternative statistics relying
upon the same principle are Fay and Wu’s H statistic (Fay
and Wu, 2000) that evaluates the number of derived
nucleotide variants at low and high frequencies with the
number of variants at intermediate frequencies; and Fu and
Li’s D, D*, F, and F* statistics (Fu and Li, 1993) that
compare the number of derived nucleotide variants observed only once in a sample with either the total number of
derived nucleotide variants (D and D*) or the mean
pairwise difference between sequences (F and F*). In
addition to allele distribution, signatures of selection can
also be evaluated by comparisons of the rates of divergence
between different classes of mutations because selection
causes a reduction in levels of nucleotide diversity. The
Hudson–Kreitman–Aguade’s HKA statistic (Hudson et al.,
1987) is the most popular example of this methodology.
HKA compares the degree of polymorphism within and
between species at two or more loci.
One of the difficulties in applying studies for the detection
of selection at specific genomic sites is the estimation of how
demographic history (population structure) affects genetic
variation in the entire genome. The inference of population
stratification can be addressed with information from
independent genetic markers but at the expense of assuming
that demographic effects occur in a similar manner across
the whole genome (Gupta et al., 2005). Population stratification should be properly corrected for a better control of
false discovery rates in tests of selection. If the allele
frequency distribution of a population is, for example,
skewed toward low-frequency variants this can be erroneously viewed as a perturbation of a standard neutral model,
thus resulting in an overestimation of the proportion of
selected genes. Wright and Gaut (2005) reviewed the most
determinant factors that seem directly to affect the capacity
of several statistical tests to detect selection. These authors
argued that, despite producing important biases, population
admixture could be corrected by means of coalescence
simulations. This technique generates samples under different null models of NE in which it is assumed that all genes
in a population are ultimately inherited from a single
common ancestor so that selection patterns can be simulated by changes in effective population size by expanding
or contracting coalescence times (Eyre-Walker et al., 1998;
Hudson, 2002). In addition to past demographic history,
the degree of LD between polymorphisms strongly affects
the sensitivity of the statistical tests to detect signatures of
selection. High levels of LD may confound true
target alleles of selection with hitchhiked alleles coupled to
another target gene because the two alleles are more likely
to be inherited together (Barton, 2000). Considering that
the extent of LD varies for species and traits there will be
more opportunities to identify a target rather than hitchhiked genes for species/traits in which LD decays very
rapidly.
Selective genes for major crops have been found in maize
(Yamasaki et al., 2005), Arabidopsis (Nordborg et al., 2005),
barley (Morrell et al., 2003), soybean (Hyten et al., 2006),
and sorghum (Hamblin et al., 2006). Notwithstanding these
remarkable examples, much still remains to be learned
about selection. One of the most important aspects to be
considered is the sampling strategy of the genes for scrutiny.
In fact, many genes are studied because they are hypothesized a priori to be under selection. To avoid overestimation
of discovery rates due to sampling biases, random genomewide surveys become critical. This type of analysis computes
statistical tests for sets of genes evenly distributed across the
genome thereby lessening false discovery rates. These
statistical methods leave unresolved the question of multiple
testing so they are still in development for humans (McVean
and Spencer, 2006) and effective progress in plants has been
restricted to maize (Wright et al., 2005).
Validation of functional polymorphisms in
heterogeneous genetic backgrounds
Allelic association refers to the relationship between a phenotypic trait and the genotype at a locus. There is a variety
of statistical methods for association mapping routinely
exploited to map genes of complex diseases in humans
(Risch, 2000), which are now largely applied in plants
(Mackay and Powell, 2007). In association mapping, LD
can be the result not only of (physical) linkage but also of
population admixture, genetic drift, and selection. The
resolution power of association mapping ultimately depends
on the structure of LD across the genome as well as how
rapidly LD decays with physical distance (Hirschhorn and
Daly, 2005). Association mapping is difficult in structured
populations, leading to spurious results if this feature is not
taken into consideration in the statistical tests. When the
functional variants are unequally distributed among different subgroups for the trait under study the association
analysis leads to false evidence for allelic association
(Knowler et al., 1988). Several facts are likely to create high
levels of population structure in very diverse panels of
individuals maintained in seed banks. First, seed bank
collections are more often than not organized in the form
of regional accessions, in some cases the accessions are
sampled in a single field-collecting trip thus accentuating
even more stratification effects. Second, plant populations
still inhabiting the wild have often had a limited gene flow,
which makes them more susceptible to population differentiation (Sharbel et al., 2000). Third, population structure
can become highly trait-dependent for wild forms and
landraces, especially for traits playing a pivotal role in local
plant adaptation, such as seed dormancy.
The first methodologies implemented for marker-trait
associations relied upon comparisons of trait mean shifts
for the different allelic states at a single locus by using
classical forms of t tests (parametric or non-parametric),
Pearson tests or Fisher exact tests (Balding, 2006). Alternative methods to control for population admixture search for
Plant molecular population genetics and seed banks | 2547
evidences of background structure and account for it
directly into the association statistic test. Genomic Control
and Structured Association, both extensively used in animal
and plant systems, are examples of these methods (Mackay
and Powell, 2007). For Genomic Control analysis a set of
random markers is used to assess the bias of the statistical
tests explained by population structure (Devlin and Roeder,
1999). The general strategy of Structured Association is first
to classify individuals into subpopulations according to the
evolutionary history of a large number of independent
genetic markers across the genome, and later to perform
marker-trait association tests within the established subgroups. Subpopulation membership has been largely explored by a popular Bayesian-based model developed by
Pritchard et al. (2000a, b) although less computationally
demanding models, i.e. genetic distance-based methods like
principal component or cluster analysis, can also be
considered for this type of analysis (Zhao et al., 2007).
Advanced statistical methods, that incorporate pedigree
relationships and population structure at the same time in
the models, have recently emerged in plant research. The
mixed-model framework offers a high degree of flexibility
for this purpose because it can account for multiple levels of
relatedness by using a genotypic relationship matrix to
structure the variance–covariance matrix between individuals. Studies in maize and potato have confirmed the value of
this approach, showing improved control of false positives
compared to classical forms of t tests performed within
prior identified subgroups (Yu et al., 2006; Malosetti et al.,
2007).
Achievements in LD mapping for plants, including
specific examples of both candidate-gene testing and
genome-wide surveys, can be found in the comprehensive
reviews of Flint-Garcia et al. (2003), Gupta et al. (2005),
and Zhu et al. (2008). Most of these studies have been
performed with highly diverse collections of annual crops,
but, recently, several cases with positive marker-trait
associations for perennial species have also been published,
such as loblolly pine (Pinus taeda) (Gonzalez-Martinez
et al., 2007), grapevine (Vitis vinifera) (This et al., 2007),
eucalyptus (E. nitens) (Thumma et al., 2005), and perennial
ryegrass (Skøt et al., 2007).
Evaluation of phenotypes across different
germplam pools
Seed banks normally possess passport data that include
taxonomy, life history, ethnobotanical knowledge or ecogeographic patterns of the collecting sites for the seed
accessions that they maintain. This basic information serves
for primary characterization and classification of the
collections. Nevertheless, the phenotypic evaluation of the
seed bank entries for potentially valuable agronomic traits
results very daunting due to the actual sizes of the whole
collections, more often than not reaching tens of thousands
of entries. In the interest of cost-effective characterization of
the plant genetic resources held in seed banks, Frankel
(1984) proposed the development of core collections; these
are subsets of the whole collection chosen as representing
most of the genetic diversity found in the collection sample.
The phenotypic screening is initially restricted to the core
collection, and if desirable phenotypes are found, then only
those accessions of the whole collection sharing similar
characteristics to the flagged individuals of the core subset
(i.e. common ecogeographic origin, genetic resemblance) are
evaluated.
Phenotypic testing of core collections largely responds,
apart from the economical obstacles or space limitations
(FAO, 1997), to the level of genetic complexity of the trait
of interest. Simple phenotypes can be directly scored over
seed lots, with no need to grow plants in the field (Fig. 3).
Direct measurements in kernels generally concern strongly
Fig. 3. Schematic diagram of the phenotypic methods used for the evaluation of agronomic traits across sets of wild relatives of crops,
landraces, and modern cultivars.
2548 | Prada
heritable characters which are largely independent of the
environment. These phenotypes usually result from the
accumulation of specific metabolites controlled by single
inherited genes, like those responsible for biosynthetic
enzymes (Doebley et al., 2006). An example of seed
attributes publicly available for applied research is the seed
information database of the Royal Botanic Gardens
Kew (http://www.kew.org/data/sid). The impact that this
class of phenotypes has on modern agriculture is, however,
very limited, and has been restricted to several traits with
explicit use to humans, such as new sources of medicines or
nutrition.
More complex phenotypes in core collections must be
measured in plant populations grown under specific field
experiments, or even indoor pots (Fig. 3). The extent to
which core collections are planted and characterized for
traits with agronomic importance is widely variable, and
mainly relates to the specific focus of each seed bank (FAO,
1997). Van Hintum et al. (2000) provide a comprehensive list
of traits for which some core collections have been screened
in the field, including, among others, various diseaseresistance traits and abiotic stress tolerance. But when
core collections are tested in the field it is important to consider that, for a target species, different populations sampled
in different habitats often exhibit local (ecotypic) adaptation
to site conditions (Schoen and Brown, 2001). Common
garden experiments are the classical designs for the analysis
of ecotypic adaptation. In this type of experiment seed lots
of two (or more) populations of the same species, but
having different geographical origins, are planted in a common environment to allow the distinction of heredity from
local adaptation. Traditionally, these designs have been
deployed for the evaluation of plant growth and plant
architecture in populations resulting from the natural
hybridization of crops and their wild relatives (Jarvis and
Hodgkin, 1999).
A series of trials containing sets of genotypes tested
across different years and locations is the method of choice
for the evaluation of those traits with high degrees of
genetic complexity, normally subjected to strong genotypeby-environment interactions (Fig. 3). These networks of
experiments have been historically managed by public or
private breeding programmes, requiring advanced field
designs and statistical methods to gain a deeper insight into
the genetic bases of the traits under study. The assessment
of phenotypic adaptation in multi-environment traits has
often relied upon empirical methods (e.g. yield per se),
based on the differential genotypic responses to environmental changes. Nevertheless, as new and more refined
methods for linkage and association mapping are emerging,
the analysis of genotype-by-environment interactions is
being moved towards the dissection of QTL-by-environment
interactions. Furthermore, advanced statistical models have
also been used as an aid to the introduction of relevant
environmental (climatic/edaphic) factors into statistical
linkage mapping models for in-depth analysis of genomic
regions that show an environmental-dependent contribution
to the phenotypes (Yin et al., 2004).
Rare alleles in seed banks: can they sensibly
impact on agronomical traits?
Alleles at low frequencies in seed bank accessions may
represent, if identified in the original pool, interesting
variants conferring local or/and wide adaptation for crop
improvement. Two questions must be properly addressed
before considering the contribution that rare alleles can
have for crop improvement. First, are rare alleles really
represented in the germplam collection? To respond to this
question many studies have focused on practical considerations of seed collection strategies (Way, 2003). In fact,
the sampling strategies seek to balance the risk of failing to
collect rare alleles against the daunting challenge of collecting very large sample sizes. Second, can the functional
variation of rare alleles be efficiently identified with the
resolution exhibited by the association mapping
approaches? This feature must be seen not only in terms of
allele frequencies but also in the proportion of individuals
that, despite having the allele in question, do not express its
phenotype.
The scope of the seed collecting strategies is to maximize
the genetic diversity sampled in the wild. Since the genetic
variation present among and within target populations is
largely unknown in advance of sampling, the general
guidelines for collecting usually rely on the analysis of
theoretical models for genetic variation in the population
sample. From theoretical breakthroughs, several population
genetic models based on molecular marker information
have been developed to assist in determining minimal
sample requirements (Crossa, 1989; Schoen and Brown,
2001). Most of these models are built around the infinite,
selectively neutral allele model of Kimura, assuming
Hardy–Weinberg equilibrium and, when possible, incorporating breeding system and population distribution (Brown
and Briggs, 1991). Brown (1989), for instance, using the
sampling theory for selectively neutral alleles showed that
the number of alleles captured in a sample was approximately proportional to the natural logarithm of its size. To
estimate the cost-effectiveness of sampling rare alleles,
Marshall and Brown (1975) classified the allelic variants
present in wild populations into four classes on the basis of
their frequencies and geographic distributions. (i) Commonwidespread alleles: they are almost certainly included even
in small samples collected from only a few populations. (ii)
Rare-widespread alleles: the target populations containing
this type of alleles behave as a single, large and unstructured
population. (iii) Common-localized alleles: they occur in
only one or a few habitats reaching a high frequency in each
of them. (iv) Rare-localized alleles: the inclusion of an allele
of this class will be unusual and serendipitous, even in very
large samples taken from a large number of populations.
To evaluate the sensitivity of the association mapping
methods to detect the effects of rare variants, both allele
class frequency and allelic effect size (allelic penetrance),
should be considered in a joint analysis. The two phenomena affect the statistical power of any association mapping
approach although they are inversely related (Morton,
Plant molecular population genetics and seed banks | 2549
Table 1. Effects that linkage disequilibrium and population structure may have on activities associated with seed banking and
identification of potentially valuable agronomic alleles in seed banks
Molecular marker-based
activity
Linkage disequilibrium (LD):
correlation among alleles
Population stratification (PS): over/underestimation
of allele frequencies
Collecting strategies maximizing
molecular marker diversity
High LD between target markers and causative
(adaptive) genes may help to capture the phenotypic
variation present in the wild
High LD between target markers and selective
genes may create hitchhiking effects in tests of
selection
Low LD can erase signals of selective
sweep so the power of the tests is diminished
Low LD in genic regions facilitates candidate-gene
testing
High LD between target markers is desirable
for genome-wide scans
High PS determined in the population sample through
marker-based estimations (i.e. Fst) must be contrasted
with phenotypic differentiation due to geographic subpopulations
High PS in the germplasm collection may confound low
allele frequencies with signals of positive selection, thus
enhancing false positives in tests of selection
Identification of selective genes
in seed bank collections
Association mapping across
highly diverse collections of
germplasm
1998). Major (Mendelian) genes having a large impact on
the phenotype are mostly present at very low frequencies in
wild populations, while polygenes with small effects on the
phenotype account mostly for all alleles underlying the
expression of a complex trait. In humans, several models
have been implemented to parameterize the combined effect
of allelic frequency and size at a single locus, either by
correcting these coupled effects in a joint pooled statistic
(Zondervan and Cardon, 2004) or by designing enriched
mating schemes to increase population size and relative
allelic contribution at the same time (Antoniou and Easton,
2003). In plant research, however, these strategies have
deserved little attention, in part because most agronomic
traits show no clear segregation patterns comparable to
those found for Mendelian disorders in humans (Hirschhorn and Daly, 2005). Risch (2000) indicated that highly
penetrant alleles with intermediate frequencies are realistically expected to be detected in association mapping,
whereas alleles present at the same frequencies but with
a more modest contribution to phenotype are practically
impossible to detect.
High PS in the germplasm collection requires association
models that correct for it, showing then a better control
of false positive discovery rates
banks. Plant molecular population genetics seems to possess
many of the scientific and technological potentials required
for such experimental designs. This area of research has
undergone drastic innovations in the last 10 years due to the
steady deposition of informative single-nucleotide polymorphisms (SNP) into large panels, partly because of the rapid
decrease of genotyping costs, and the ongoing improvements in algorithms for sequence data analyses (i.e. more
refined methods to deal with population structure and
better characterization of background LD patterns). The
deployment of these techniques with increasing rates of
interconnection holds real promise for a better understanding on how genetic variation is arrayed in modern crops and
wild forms. Far too often genetic research for these two
plant genetic backgrounds has been undertaken separately
by the scientific community.
Acknowledgements
The author thanks Dr JPA Heuts (Eindhoven University of
Technology) and two anonymous reviewers for very valuable comments.
Conclusions and future prospects
Major genetic divergences between ‘wild’ and ‘cultivated’
genetic pools are largely explained by human-mediated
selection through domestication, founding events, and
breeding practices. Theses processes have presumably
created different genetic bottlenecks which have resulted in
decreasing rates of genetic diversity, changes in allele
frequencies, increases in LD, and reduction of rare alleles
in modern crops (Halliburton, 2004). Therefore any discovery initiative for the characterization of agronomically
related alleles in seed banks should be much affected by
analogous features, such as population stratification, linkage disequilibrium, sample size, allelic penetrance, and allele
frequency distribution (Table 1). There are consequently
many small, but crucially important, practical and statistical
choices that have to be made for good experimental designs
in studies with wild forms and landraces maintained in seed
References
Antoniou AC, Easton DF. 2003. Polygenic inheritance of breast
cancer: implications for design of association studies. Genetic
Epidemiology 25, 190–202.
Balding DJ. 2006. A tutorial on statistical methods for population
association studies. Nature Reviews Genetics 7, 781–791.
Barnaud A, Lacombe T, Doligez A. 2006. Linkage disequilibrium in
cultivated grapevine, Vitis vinifera L. Theoretical and Applied Genetics
112, 708–716.
Barton N. 2000. Genetic hitchhiking. Philosophical Transactions of
the Royal Society B, Biological Sciences 355, 1553–1562.
Brown AHD. 1989. The case for core collections. In: Brown AHD,
Frankel OH, Marshall DR, Williams JT, eds. The use of plant genetic
resources. Cambridge, UK: Cambridge University Press, 136–156.
2550 | Prada
Brown AHD, Briggs JD. 1991. Sampling strategies for genetic
variation in ex situ collections of endangered plant species. In: Falk DA,
Holsinger KE, eds. Genetics and conservation of rare plants. New
York, NY: Oxford University Press, 99–123.
Forrester JW. 1961. Industrial dynamics. Cambridge, MA: MIT Press.
Frankel OH. 1984. Genetic perspectives of germplasm conservation.
In: Arber W, Llimensee K, Peacock WJ, Starlinger P, eds. Genetic
manipulation: impact on man and society. Cambridge, UK: Cambridge
University Press, 161–170.
Brown GR, Gill GP, Kuntz RJ, Langley CH, Neale DB. 2004.
Nucleotide diversity and linkage disequilibrium in loblolly pine. Proceedings of the National Academy of Sciences, USA 42, 15255–15260.
Frary A, Nesbitt TC, Frary A, et al. 2000. fw22: a quantitative trait
locus key to the evolution of tomato fruit size. Science 289, 85–88.
Buckler ES, Thornsberry JM, Kresovich S. 2001. Molecular diversity,
structure and domestication of grasses. Genetics Research 77, 213–218.
Fu YX, Li WH. 1993. Statistical tests of neutrality of mutations.
Genetics 133, 693–709.
Caldwell KS, Russell J, Langridge P, Powell W. 2006. Extreme
population-dependent linkage disequilibrium detected in an inbreeding
plant species, Hordeum vulgare. Genetics 172, 557–567.
Gonzalez-Martinez SC, Wheeler NC, Ersoz E, Nelson CD,
Neale DB. 2007. Association genetics in Pinus taeda L.I. Wood
property traits. Genetics 175, 399–409.
Camus-Kulandaivelu L, Veyrieras J-B, Madur D, Combes V,
Fourmann M, Barraud S, Dubreuil P, Gouesnard B, Manicacci D,
Charcosset A. 2006. Maize adaptation to temperate climate: relationship between population structure and polymorphism in the
Dwarf8 gene. Genetics 172, 2449–2463.
Grandillo S, Ku HM, Tanksley SD. 1999. Identifying the loci
responsible for natural variation in fruit size and shape in tomato.
Theoretical and Applied Genetics 99, 978–987.
Crossa J. 1989. Methodologies for estimating the sample size
required for genetic conservation of outbreeding crops. Theoretical
and Applied Genetics 77, 153–161.
Devlin B, Roeder K. 1999. Genomic control for association studies.
Biometrics 55, 997–1004.
Doebley J, Stec A, Gustus C. 1995. teosinte branched1 and the
origin of maize: evidence for epistasis and the evolution of dominance.
Genetics 141, 333–346.
Doebley J, Stec A, Hubbard L. 1997. The evolution of apical
dominance in maize. Nature 386, 485–488.
Doebley JF, Gaut BS, Smith BD. 2006. The molecular genetics of
crop domestication. Cell 127, 1309–1321.
Doney DL, Whitney ED. 1990. Genetic enhancement in Beta for
disease resistance using wild relatives: a strong case for the value of
genetic conservation. Economic Botany 44, 445–451.
Dvornyk V, Sirviö A, Mikkonen M, Savolainen O. 2002. Low
nucleotide diversity at the pal1 locus in the widely distributed Pinus
sylvestris. Molecular Biology and Evolution 19, 179–188.
Esquinas-Alcazar J. 2005. Protecting crop genetic diversity for food
security: political, ethical and technical challenges. Nature Reviews
Genetics 6, 946–953.
Eyre-Walker A, Gaut RL, Hilton H, Feldman DL, Gaut BS. 1998.
Investigation of the bottleneck leading to the domestication of maize.
Proceedings of the National Academy of Sciences, USA 95, 4441–4446.
FAO. 1997. The state of the world’s plant genetic resources for food
and agriculture. http://www.fao.org/WAICENT/FAOINFO/AGRICULT/
AGP/AGPS/Pgrfa/pdf/swrfull.pdf.
Fay JC, Wu CI. 2000. Hitchhiking under positive Darwinian selection.
Genetics 155, 1405–1413.
Flint-Garcia SA, Thornsberry JM, Buckler ES. 2003. Structure of
linkage disequilibrium in plants. Annual Reviews in Plant Biology 54,
357–374.
Flint-Garcia SA, Thuillet AC, Romero SM, Mitchell S, Doebley J,
Kresovich S, Goodman MM, Buckler ES. 2005. Maize association
population: a high resolution platform for QTL dissection. The Plant
Journal 44, 1054–1064.
Gupta PK, Rustgi S, Kulwal PL. 2005. Linkage disequilibrium and
association studies in higher plants: present status and future
prospects. Plant Molecular Biology 57, 461–485.
Halliburton R. 2004. Introduction to population genetics. Upper
Saddle River, NJ: Pearson-Prentice-Hall.
Hamblin MT, Casa AM, Sun H, Murray SC, Paterson AH,
Aquadro CF, Kresovich S. 2006. Challenges of detecting directional
selection after a bottleneck: lessons from Sorghum bicolor. Genetics
173, 953–964.
Harlan JR. 1975. Crops and man. Madison, WI: American Society of
Agronomy.
Hirschhorn JN, Daly MJ. 2005. Genome-wide association studies for
common diseases and complex traits. Nature Reviews Genetics 6, 95–108.
Hoisington D, Khairallah M, Reeves T, Ribaut J-M, Skovmand B,
Taba S, Warburton M. 1999. Plant genetic resources: what can they
contribute toward increased crop productivity? Proceedings of the
National Academy of Sciences, USA 96, 5937–5943.
Hudson RR. 2002. Generating samples under a Wright–Fisher neutral
model of genetic variation. Bioinformatics 18, 337–338.
Hudson RR, Kreitman M, Aguade M. 1987. A test of neutral
molecular evolution based on nucleotide data. Genetics 116, 153–159.
Hyten DL, Song Q, Zhu Y, Choi I, Nelson RL, Costa JM,
Specht JE, Shoemaker RC, Perry BC. 2006. Impacts of genetic
bottlenecks on soybean genome diversity. Proceedings of the National
Academy of Sciences, USA 103, 16666–16671.
Ingvarsson PK. 2005. Nucleotide polymorphism and linkage disequilibrium within and among natural populations of European aspen
(Populus tremula L., Salicaceae). Genetics 169, 945–953.
Jarvis DI, Hodgkin T. 1999. Wild relatives and crop cultivars:
detecting natural introgression and farmer selection of new genetic
combinations in agroecosystems. Molecular Ecology 8, 159–173.
Johnson R. 2004. Marker-assisted selection. Plant Breeding Reviews
24, 293–309.
Jones H, Leigh FJ, Mackay I, Bower MA, Smith LMJ,
Charles MP, Jones G, Jones MK, Brown TA, Powell W. 2008.
Population-based resequencing reveals that the flowering time adaptation of cultivated barley originated east of the Fertile Crescent.
Molecular Biology and Evolution 25, 2211–2219.
Plant molecular population genetics and seed banks | 2551
Kimura M. 1968. Evolutionary rate at the molecular level. Nature 217,
624–626.
Knowler WC, Williams RC, Pettitt DJ, Steinberg AG. 1988.
Gm3;5,13,14 and type 2 diabetes mellitus: an association in American
Indians with genetic admixture. The American Journal of Human
Genetics 43, 520–526.
Krutovsky KV, Neale DB. 2005. Nucleotide diversity and linkage
disequilibrium in cold-hardiness- and wood quality-related candidate
genes in Douglas fir. Genetics 171, 2029–2041.
Mackay I, Powell W. 2007. Methods for linkage disequilibrium
mapping in crops. Trends in Plant Science 12, 57–63.
Malosetti M, van der Linden CG, Vosman B, van Eeuwijk FA.
2007. A mixed-model approach to association mapping using
pedigree information with an illustration of resistance to Phytophthora
infestans in potato. Genetics 175, 879–889.
Marshall DR, Brown AHD. 1975. Optimum sampling strategies in
genetic conservation. In: Frankel OH, Hawkes JG, eds. Crop genetic
resources for today and tomorrow. Cambridge, UK: Cambridge
University Press, 53–80.
McVean G, Spencer CCA. 2006. Scanning the human genome for
signals of selection. Current Opinion in Genetics and Development 16,
624–629.
Morrell PL, Lundy KE, Clegg MT. 2003. Distinct geographic
patterns of genetic diversity are maintained in wild barley (Hordeum
vulgare ssp. spontaneum) despite migration. Proceedings of the
National Academy of Sciences, USA 100, 10812–10817.
Morton NE. 1998. Significance levels in complex inheritance. The
American Journal of Human Genetics 62, 690–697.
Nordborg M, Hu TT, Ishino Y, et al. 2005. The pattern of
polymorphism in Arabidopsis thaliana. Public Library of Sciences
Biology 3, e196.
Palaisa K, Morgante M, Tingey S, Rafalski A. 2004. Long-range
patterns of diversity and linkage disequilibrium surrounding the maize
Y1 gene are indicative of an asymmetric selective sweep.
Proceedings of the National Academy of Sciences, USA 101,
9885–9890.
Parmesan C, Yohe G. 2003. A globally coherent fingerprint of climate
change impacts across natural systems. Nature 421, 37–42.
Paterson AH, Damon S, Hewitt JD, Zamir D, Rabinowitch HD,
Lincoln ES, Lander ES, Tanksley SD. 1991. Mendelian factors
underlying quantitative traits in tomato: comparison across species,
generations and environments. Genetics 127, 181–197.
Pritchard JK, Stephens M, Donnelly P. 2000a. Inference of
population structure using multilocus genotype data. Genetics 155,
945–959.
Pritchard JK, Stephens M, Rosenberg NA, Donnelly P. 2000b.
Association mapping in structured populations. The American Journal
of Human Genetics 67, 170–181.
Putterill J, Laurie R, Macknight R. 2004. It’s time to flower: the
genetic control of flowering time. BioEssays 26, 363–373.
Rajaram S, Mann CE, Ortiz-Ferrara G, Mujeeb-Kazi A. 1983.
Adaptation, stability and high yield potential of certain 1B/1R CIMMYT
wheats. In: Sakamoto S, ed. Proceedings of the 6th international
wheat genetics symposium. Kyoto: Japan, 613–621.
Risch NJ. 2000. Searching for genetic determinants in the new
millennium. Nature 405, 847–856.
Salamini F, Ozkan H, Brandolini A, Schafer R, Martin W. 2005.
Genetics and geography of wild cereal domestication in the near east.
Nature Reviews Genetics 3, 429–441.
Savitsky H. 1960. Meiosis in an F1 hybrid between a Turkish wild beet
(Beta vulgaris ssp. maritima) and Beta procumbens. Journal of the
American Society for Sugar Beet Technology 11, 49–67.
Sawers RJH, Sheehan MJ, Brutnell TP. 2005. Cereal phytochromes: targets of selection, targets for manipulation? Trends in Plant
Sciences 10, 138–143.
Schoen DJ, Brown AHD. 2001. The conservation of wild plant
species in seed banks. BioScience 51, 960–966.
Sharbel TF, Haubold B, Mitchell-Olds T. 2000. Genetic isolation by
distance in Arabidopsis thaliana: biogeography and postglacial colonization of Europe. Molecular Ecology 9, 2109–2118.
Skøt L, Humphreys J, Humphreys MO, Thorogood D,
Gallagher J, Sanderson R, Armstead IP, Thomas ID. 2007.
Association of candidate genes with flowering time and water-soluble
carbohydrate content in Lolium perenne (L.). Genetics 177, 535–547.
Tajima F. 1989. Statistical method for testing the neutral mutation
hypothesis by DNA polymorphism. Genetics 123, 585–595.
Tanksley SD, McCouch SR. 1997. Seed banks and molecular
maps: unlocking genetic potential from the wild. Science 277,
1063–1066.
This P, Lacombe T, Cadle-Davidson M, Owens CL. 2007. Wine
grape (Vitis vinifera L.) color associates with allelic variation in the
domestication gene VvmybA1. Theoretical and Applied Genetics 114,
723–730.
Peng J, Richards DE, Hartley NM, et al. 1999. ‘Green revolution’
genes encode mutant gibberellin response modulators. Nature 400,
256–261.
Thornsberry JM, Goodman MM, Doebley J, Kresovich S,
Nielsen D, Buckler ES. 2001. Dwarf8 polymorphisms associate with
variation in flowering time. Nature Genetics 28, 286–289.
Piffanelli P, Ramsay L, Waugh R, Benabdelmouna A, D’Hont A,
Hollricher K, Jorgensen JH, Schulze-Lefert P, Panstruga R.
2004. A barley cultivation- associated polymorphism conveys resistance to powdery mildew. Nature 430, 887–891.
Thumma BR, Nolan MF, Evans R, Moran GF. 2005. Polymorphisms in Cinnamoyl CoA Reductase (CCR) are associated with
variation in microfibril angle in Eucalyptus spp. Genetics 171,
1257–1265.
Ponting RC, Drayton MC, Cogan NOI, Dobrowolski MP,
Spangenberg GC, Smith KF, Forster JW. 2007. SNP discovery,
validation, haplotype structure and linkage disequilibrium in full-length
herbage nutritive quality genes of perennial ryegrass (Lolium perenne
L.). Molecular Genetics and Genomics 278, 585–597.
van Hintum TJL, Brown AHD, Spillane C, Hodgkin T. 2000. Core
collections of plant genetic resources. IPGRI Technical Bulletin No 3.
Rome, Italy: International Plant Genetic Resources Institute.
Way MJ. 2003. Collecting seed from non-domesticated plants for
long-term conservation. In: Smith RD, Dickie JB, Linington SH,
2552 | Prada
Pritchard HW, Probert RJ, eds. Seed conservation: turning science
into practice. London, UK: Royal Botanic Gardens Kew, 163–201.
to the Arabidopsis flowering time gene CONSTANS. The Plant Cell 12,
2473–2483.
Weber A, Clark RM, Vaughn L, Sanchez-Gonzalez J-J, Yu J,
Yandell BS, Bradbury P, Doebley J. 2007. Major regulatory genes
in maize contribute to standing variation in teosinte (Zea mays ssp.
parviglumis). Genetics 177, 2349–2359.
Yin X, Struik PC, Kropff MJ. 2004. Role of crop physiology in
predicting gene-to-phenotype relationships. Trends in Plant Science 9,
426–432.
Wright SI, Gaut BS. 2005. Molecular population genetics and the
search for adaptive evolution in plants. Molecular Biology and
Evolution 22, 506–519.
Wright SI, Vroh-Bi I, Schroeder SG, Yamasaki M, Doebley JF,
McMullen MD, Gaut BS. 2005. The effects of artificial selection on
the maize genome. Science 308, 1310–1314.
Yamasaki M, Tenaillon MI, Vroh-Bi I, Schroeder S, SanchezVilleda H, Doebley J, Gaut BS, McMullen MD. 2005. A large-scale
screen for artificial selection in maize identifies candidate agronomic
loci for domestication and crop improvement. The Plant Cell 17,
2859–2872.
Yano M, Katayose Y, Ashikari M, et al. 2000. Hd1, a major
photoperiod sensitivity quantitative trait locus in rice, is closely related
Yu J, Pressoir G, Briggs WH, et al. 2006. A unified mixed-model
method for association mapping that accounts for multiple levels of
relatedness. Nature Genetics 38, 203–208.
Yu J, Buckler ES. 2006. Genetic association mapping and genome
organization of maize. Current Opinion in Biotechnology 17,
155–160.
Zhao K, Aranzana MJ, Kim S, et al. 2007. An Arabidopsis example
of association mapping in structured samples. Public Library of
Sciences Genetics 3, e4.
Zhu C, Gore M, Buckler ES, Yu J. 2008. Status and prospects of
association mapping in plants. The Plant Genome 1, 5–20.
Zondervan KT, Cardon LR. 2004. The complex interplay among
factors that influence allelic association. Nature Reviews Genetics 5,
89–100.