Agriculture Reference
In-Depth Information
cated that fertilizer use has leveled off in the
Yaqui Valley while yields are still increasing, but
perhaps at a reduced rate (Reynolds et al., 1999;
Rajaram and Braun 2008). Reynolds et al. (1999)
attributed this to an increase in N-use effi ciency
by new cultivars, indicating that the genetic
component of grain yield improvement has
increased.
Schmidt (1984) used yield data obtained from
uniform regional nurseries in the US to deter-
mine genetic gain from 1958 to 1980. In compari-
son to the long-term check cultivars used in the
nurseries, the yield advantage of the top experi-
mental lines increased from 25% in 1959 to 46%
in 1979, an annual rate of gain of 0.74%. He then
compared this genetic gain with the on-farm
wheat yields in the US across the same time
period and determined that approximately one-
half of on-farm yield gains could be attributed to
breeding for yield. Brancourt-Hulmel et al. (2003)
reported that between 33% and 63% of the yield
gains in France were genetically driven.
Wheat grain yield in the UK increased linearly
by 110 kg ha
−1
yr
−1
from 1950 to 1995, but N use
peaked around 1985 (Austin 1999). The contin-
ued increase in grain yield after 1985, without an
increase in N, implies that genetic improvement
played a vital role in later years. Figure 17.1 illus-
trates that yield increases in the UK have not
always been constant. It appears that growth was
relatively slow from 1961 through 1976, followed
by a period of rapid growth from 1976 to 1996,
and no growth from 1996 to 2007. It is encourag-
ing that Shearman et al. (2005) showed that
genetic gain in the UK is still occurring. Cultivar
replacement and management changes take place
over several years, and new cultivars may not yet
be fully adopted. Calderini and Slafer (1998) ana-
lyzed yield trends in 21 countries and found that
gains were often bilinear or trilinear, having dif-
ferent slopes over different periods of years.
These qualitative changes were likely related to
major shifts in cultivars or management
practices.
Based on the type of evidence just cited, it is
generally projected that 50% of the increase in
wheat grain yield is due to the adoption of new
cultivars (genetic), while the other 50% is due to
improved management practices. In reality, these
are hard to separate empirically or theoretically.
Borlaug (2007) stressed that the Green Revolu-
tion was not the result of introducing high-yield-
ing semidwarf wheat cultivars or N fertilizers
alone but the product of plant breeding and agro-
nomic practices working in combination.
EMPIRICAL ESTIMATION OF
GENETIC GAIN
Numerous authors have reviewed genetic gains in
wheat, including Austin et al. (1989), Calderini
and Slafer (1998), Austin (1999), Reynolds et al.
(1999), Brancourt-Hulmel et al. (2003), Foulkes
et al. (2007), and Rajaram and Braun (2008).
These authors point to diffi culties in quantifying
genetic gain independently of crop management
changes, but all agree that genetic gain estimates
are needed to detect any degree of breeding prog-
ress over time. There are two methods commonly
used to estimate genetic gains. One is to examine
wheat cultivar performance trials over a period of
time and compare new cultivar performance with
a common control cultivar; examples are Schmidt
(1984) and Donmez et al. (2001). Another method
is to compare old and new cultivars in common
yield trials (Cox et al., 1988; Brancourt-Hulmel
et al., 2003; Shearman et al., 2005). Cox et al.
(1988) describe this method as “a direct compari-
son of old and new cultivars, but (taking) place on
the new cultivars' turf.”
Table 17.2 summarizes 18 representative
studies from 11 countries. Not surprisingly,
newer wheat cultivars yielded more than older
cultivars, but the rates of gain and the associated
yield components varied greatly.
Grain yield
The data summarized in Table 17.2 reveal a wide
range in estimates of genetic gain for grain yield
among environments. Most data were collected in
multiple natural or modifi ed environments. The
natural environments were used to simulate on-
farm results. The input levels in these experi-
ments were similar to what was commonly used
