Improvement of Yield per se in Sugarcane - Translational Genomics for Crop Breeding: Abiotic Stress, Yield and Quality - page 224

of potential additional gains for sucrose content
within 'noble' species and illustrate the potential
advantage of using MAS approaches to purge
the domesticated species of alleles that reduce
sucrose content.

on the thresholds used to qualify an association
as statistically significant. Like QTL studies of
interspecific crosses (Sills et al. 1995; Ming et al.
2002b; Alwala et al. 2009), some modern QTL
studies revealed 'significant' digenic interactions
(Hoarau et al. 2002; Aitken et al. 2006; Aitken
et al. 2008; Pinto et al. 2010). These digenic inter-
actions are responsible for a substantial portion
of phenotypic variance, illustrating the poten-
tial importance of epistasis in the genetic control
of yield components. In several QTL studies,
populations were phenotyped in successive crop
cycles (Hoarau et al. 2002; Reffay et al. 2005;
Piperidis et al. 2008; Pinto et al. 2010) and some-
times in several locations (Jordan et al. 2004;
Aitken et al. 2006; Aitken et al. 2008; Pastina
et al. 2012). Although Aitken and colleagues
(2006) combined data across locations for sta-
tistical analysis, in most of these studies, detec-
tion of marker-trait associations was carried out
separately for each crop cycle and environment.
They usually revealed a small number of overlap-
ping sets of QTLs, however the direction of the
marker effect on the trait value (either positive
or negative) may always be conserved (Hoarau
et al. 2002). These findings illustrate the fact that
detecting QTLs in modern cultivars is highly sen-
sitive to the effects of statistical thresholds. Only
one work, recently published by Pastina and col-
leagues (2012), describes the use of mixed mod-
els for detection of yield QTLs, revealing sig-
nificant interactions for all traits for QTL x crop
cycle, QTL x environment, and QTL x crop cycle
x environment. Piperidis and colleagues (2008)
compared the location of QTLs for brix across
four modern cultivar maps (R570, MQ77-340,
Q117, and Q165), based on the use of a few neu-
tral SSR primers scattered throughout the eight
homo(eo)logy groups (HG). Two of the eight
HGs were seen to contain marker-trait associa-
tions for brix, in two or three out of the four maps,
suggesting common loci of interest in these
HGs among cultivars. These results illustrate the
value of conducting meta-QTL analyses to reveal
key alleles that could be targeted using MAS
approaches in genetic improvement program.

ModernCultivarCrosses

The second category of QTL studies was based
on five crosses involving modern cultivars (Table
13.1): selfing of R570 (Hoarau et al. 2002),
SP80-180 SP80-4966 (Da Silva and Bressiani

2005; Pinto et al. 2010; Pastina et al. 2012),
Q117

×

×
MQ77-340 (Reffay et al. 2005; Piperidis et al.
2008), and a cross between the S. officinarum
clone IJ76-514 and the modern cultivar Q165
(Aitken et al. 2006; Aitken et al. 2008). All these
studies used classical measurements of yield
components such as height, diameter, weight or
number of stalks, and measurements of the qual-
ity of sugarcane juice (brix and pol). Three of
these studies based on the largest population
size (230 to 295 individuals) revealed numer-
ous QTLs for each trait with small individual
R 2 values usually ranging between 3% and 8%
(Hoarau et al. 2002; Reffay et al. 2005; Aitken
et al. 2006; Aitken et al. 2008; Piperidis et al.
2008). These R 2 values are lower than values
found in previous works by Ming and colleagues
(Ming et al. 2001; Ming et al. 2002a; Ming et al.
2002b) for F1 interspecific populations of similar
size (239 and 264 individuals, respectively) with
R 2 that were up to 23%. These findings show
that the frequency of alleles for major effects for
yield-related traits is lower in modern varieties.
This could be due to the fact that: (1) the most
unfavorable alleles in the S. spontaneum part of
the modern sugarcane genome are likely to have
been eliminated by 'nobilization,' and (2) some
of the most favorable alleles have already been
'fixed' in several copies by recurrent selection,
which makes their detection less likely (absence
of visible segregation). As a consequence of their
relatively modest effect size, the number of QTLs
detected in modern cultivars mainly depended

× 74C42 (Jordan et al. 2004), Q117