Agriculture Reference
In-Depth Information
These long-term trials were conducted in an area
that received the same management practices
from year to year. In this case, winter wheat was
grown under conventional tillage, planted in
October, and harvested between June and July.
Evident from this work is that the demand for
fertilizer N changed dramatically each year and
this demand was unpredictable, similar to maize
results reported by Miao et al. (2006). The vari-
able demand for fertilizer N represents the infl u-
ence of the environment (rainfall and temperature),
soil type, and previous crop management on
yield potential, which in turn can affect N respon-
siveness. Factors which impact yield potential
are numerous, unpredictable, and change each
year. This could include variable plant stand
due to moisture stress, delayed planting due to
excessive autumn rainfall, surface soil crusting,
in-season moisture stress, diseases (various),
lodging, soil type, and previous history of fertil-
izer use.
From the same data presented in Fig. 10.1, the
optimum mean fertilizer N rate that resulted
in maximum wheat grain yields from 1971 to
2006 was 56 kg N ha
−1
(±42 kg N ha
−1
), and the
optimum rate varied from 1 to 156 kg N ha
−1
(Fig. 10.2). Thus, fertilizing based on the mean
optimum rate was only appropriate in 6 of 35
years, or 17% of the time. Oklahoma State Uni-
versity has numerous long-term experiments
which document fertilizer N, P, and K response
as a function of time in winter wheat (Raun
2008a). All of the long-term winter wheat trials
scattered across the state of Oklahoma confi rmed
this observation that optimum N rates change
drastically from year to year.
If the optimum N rates are known to change
signifi cantly both spatially and temporally, it is
intuitive that yield levels, and yield potential, will
change accordingly. Hence Raun et al. (2002,
2005) focused on the development of algorithms
that could predict yield potential and determine
N rates based on projected removal. The central
component of this approach is recognizing the
need to predict yield potential from sensor
readings based on midseason NDVI. While
straightforward, it is complicated by the fact that
N responsiveness, or the response index (RI),
must be determined separately. In each fi eld,
N is applied in one strip at a rate where N will
not be limiting through the season. While it is
recognized that this method of application is
highly ineffi cient, it is only being used in one
portion of the fi eld. In this regard, it is important
to recognize that N responsiveness is independent
of yield potential. Furthermore, although N
responsiveness like yield potential is dependent
upon spatial and temporal variability, the same
variables (total rainfall) can impact RI and
YP
0
differently, where YP
0
is the yield potential
that can be achieved with no added fertilizer
applied.
180
160
140
120
100
80
60
40
20
Fig. 10.2
Optimum fertilizer
N rate, by year, from long-
term Experiment 502 and the
overall mean from 1971 to
2006, at Lahoma, Oklahoma.
0
