This report documents the last two objectives of the SWEEP/TED project "Management
of Farm Field Variability". The objectives of this part of the project were
to determine crop yield response to different tillage systems, at different
landscape positions, and relate the variations in crop yield response to
soil and landform properties. The possibility of implementing variable management
was examined by quantifying the stability of the yield response pattern
over a number of years.
The study was carried out in cooperation with the provincial Tillage-2000
project, which is a cooperative project of the Ontario Ministry of Agriculture
and Food (OMAF), the Dept. of Land Resource Science (Univ. of Guelph), and
cooperating farmers. Details of the study are given in the annual reports
of the project, which are available from OMAF.
Field sites were selected from the existing Tillage-2000 sites for detailed
examination of the relationships between landform, soil properties, soil
loss, tillage management, and crop yield. At each field site, two tillage
systems (conventional, conservation) had been established and permanent
benchmark monitoring locations selected. The conservation system was defined
as any tillage system which should decrease the loss of soil from the field
by erosion. The tillage system were classified as no-till if no primary
or secondary tillage was completed, minimum till systems if no inversion
tillage (moldboard plough) system was used and moldboard if a moldboard
plough was used to invert the soil.
Landform at each benchmark was characterised by carrying out a detailed
elevation survey using a laser theodolite, and calculating parameters based
on a digital elevation model and classification system. The landform classification
is based on the slope gradient and change in slope gradient with distance
(curvature) at a location. A concave surface with low slope is classified
as a footslope while a convex surface with low slope is a crest/shoulder
position. In addition the cross-slope curvature can be causing water to
converge (concave cross-slope) or diverge (convex cross-slope) at a point
and landforms are further divided by this classification.
Relative soil loss was obtained from relative changes in 137Cs
content (a naturally occurring soil tracer) measurements. Soil properties
were obtained from a 1.0 m soil core taken at each site using a hydraulic
soil sampler. Yield measurements collected as part of the Tillage-2000 program
were entered into the data base containing all of the other information.
Soil sampling for soil fertility analysis was also completed. Indices for
plant-available water and air filled porosity were calculated from soil
profile data. The objective of the sampling program was to obtain as complete
a database as possible. A copy of the database accompanies this report.
The 5 year (1986-1990) average grain corn, soybean, and small grain yields
in the paired conservation and conventional tillage systems were not significantly
different. In addition, field- by-field, and year-to-year differences in
yield were equally variable in the conservation and conventional systems.
The paired moldboard and minimum tillage sites suggested a slight yield
advantage to the moldboard system. Average relative crop yields across all
crop types were higher in the moldboard system for all years except 1988.
A total of 23 out of 36 plot years had higher yields on the moldboard system
compared to the minimum tillage system.
The paired no-till and moldboard tillage sites indicated no significant
difference in yield for any of the crop types. A total of 14 out of 23 plot-years
had a higher yield in the no-till system than in the moldboard system. Average
relative yield across all crops was higher in 1987 and 1988 in the no-till
compared to moldboard, but lower in the other three years.
The paired no-till and minimum tillage sites indicated a slight yield
advantage to the no-till system. The average no-till yield was significantly
higher than minimum tillage yield (0.1 probability) for both the soybeans
and small grain crops.
Conservation tillage systems, both overall and in the paired tillage
comparisons, did relatively better than conventional systems in 1988 (a
drought stress year) than in any other year. This suggests that conservation
tillage systems are more buffered against adverse crop growth conditions
than conventional systems.
Soil texture (Ap horizon) class and tillage × texture class interactions
were significant (0.001 probability) in explaining the variations in yield
of all crop types. Landform class was significant for corn and soybeans,
but not small grains. A regression of the ratio of no-till yield/conventional
tillage yield against % sand content of the Ap horizon was significant (0.05
probability). The regression indicated that for % sand contents greater
than 36% the no-till yield would on average be higher than the conventional
tillage system. The no-till yield for finer textured soils would on average
be lower than conventional tillage systems. The interaction of tillage and
texture is suggested as the reason for little overall differences in tillage
system on crop yield, across all farm sites.
Landform classification criteria as proposed by Pennock and de Jong (1987)
are not sensitive enough at the elevation measurement scale used in this
study (i.e. 10 m grid). The criteria for backslope versus level class (6.5%
slope) resulted in an unacceptably large number of level class designations.
The criteria was changed to 3% slope gradient.
A general analysis of covariance confirmed the presence of significant
covariance between the major soil properties. The 7 unit landform classification
system was not very useful in explaining variations in soil properties.
This was partly attributed to the high variability of texture within the
landform classes, and the covariance of texture and other soil properties.
A sub-classification by upper, middle, and lower slope position within
each of the original landform classes explained a significant (0.05 probability)
amount of the variability of almost all soil properties, especially those
related to soil loss. The conceptual model of the slope location of the
original landform classes was correct for approximately 60-70% of the benchmarks.
However, a significant number of benchmarks were found at slope positions
not conforming to the conceptual model. The poor success of the original
7 landform classes in explaining the spatial variations of soil properties
was attributed to the conflicting patterns that water flow and tillage translocation
would have on soil properties.
Solum depth was significantly related to soil texture class, but
137Cs content was not related to texture. This suggests that the amount
of soil loss is not related to texture, and the increased solum depths in
the sandy soils are related to pedogenic processes. The independence of
soil loss on soil texture, and the higher cesium loss on upper slope compared
to middle slope positions are contrary to water erosion theory, but consistent
with tillage translocation theory.
The difference between the highest and lowest yielding areas in a field,
across all years and all sites was 40% of the mean field yield. Tillage
system did not significantly affect the within field variations in yields.
The lowest, highest, and range, of relative yield difference were similar
in the paired tillage comparisons. The standard deviation of relative yield
difference was slightly higher (0.05 probability) in the minimum compared
to no-till system.
The percentage of within field variation of yield remained constant from
year to year was on average 52% and 56% (significantly different at 0.10
probability), for the conventional and conservation tillage systems respectively.
Approximately 3% of the benchmarks had an average yield which was at least
30% less than the average field yield. Another 5% of the benchmarks were
between 20% and 30% lower than the field average yield. A total of 15%,
and 4% of the benchmarks had an average yield which were greater than 10%,
and greater than 20% of the field average yield respectively. The relative
ranking of the benchmarks with respect to relative yield was independent
of the measurement year, but the year did affect the magnitude in all yield
classes. The drought stress year (1988) resulted in much greater relative
within field variations of crop yield.
A paired benchmark analysis on crop response on stress (low yielding)
and non-stress (high yielding) benchmarks, in stress and non-stress growing
conditions was carried out. It indicated that conservation tillage systems
may be more buffered against adverse climatic growing conditions than conventional
tillage systems. High yielding areas under conservation tillage dropped
only 13.8% in yield during a stress year, compared to 16.5% decrease in
the conventional system. Low yielding areas in the conservation system decreased
24% in yield in the stress year, compared to a 31.1% decrease in the conventional
system. This benchmark data supported the paired field yield data, which
indicated that the ratio of conservation yield/conventional yield was the
highest in 1988 the drought stress year.
The 7 unit landform classification explained a significant amount of
the within field benchmark yield data. Converging landforms had on average
8.5%, 7.8%, and 5.5% higher yields of corn, soybeans, and small grains respectively,
compared to diverging landforms. Across all crop types the average difference
between converging and diverging landforms was 7.3% (significant at 0.05
probability level). The diverging shoulder and backslope position had significantly
lower corn and soybean yields, than the other landform units.
There was a significant interaction of soil texture class and soil loss,
on relative crop yield losses. Benchmarks were separated on the basis of
% sand in the Ap horizon and yield response correlated against cesium content
in each texture class. Yield response to soil loss was predicted from these
correlations. Severely eroded soils with a % sand content greater than 70%
had an average predicted yield loss of 37% of the field average yield. The
same severely eroded soils had a predicted yield loss of only 8.0%, 4.7%,
and 0.7% for Ap horizons with 50-70%, 40-50%, and 30-40% sand content. The
yield loss in all texture groupings increased during the 1988 growing season
indicating soil loss was affecting available water, but more so in the sandier
soils. The available water index of McBride and Mackintosh (1984) was significantly
related to cesium content in the > 70% sand content group, but not in the
other groups. The benchmarks with 20-30%, and < 20% sand content had a predicted
relative yield decrease of 4.3% and 8.0% respectively. Thus, the benchmarks
in the medium texture classes had much less predicted yield loss from soil
loss, than benchmarks with lighter or finer soil textures. This is consistent
with the higher available water holding capabilities of medium textured
soils compared to other textures.
The data in this project, and in the Tillage-2000 project indicate that
it should be possible to implement a conservation tillage system with no
loss in yield productivity. This is especially true for sandy textured soils
where increases in yield are likely under conservation tillage.
For the range in soil and landform conditions in this report, there seems
to be little benefit to adopting a minimum tillage system. No-till yields
were equal or better than minimum till and the minimum till was slightly
lower yielding than the moldboard tillage. In addition, the work on tillage
translocation suggests that secondary tillage, which can sometimes be more
intensive under minimum tillage, can result in significant soil loss off
of shoulder/crest slope positions. The criteria of 20% or 30% residue cover
to control water erosion is meaningless with regards to tillage translocation.
Thus, unless a management procedure can be devised to control tillage translocation,
minimum tillage can not be viewed as a viable alternative to a moldboard
system in a sustainable production system. The minimum till system may control
water and soil loss by water erosion, but it will not be stopping the decline
in soil quality and crop productivity in the upland regions.
The study indicates that the sandy soils (> 70% sand) are the most fragile
of all of the soil types. Soil loss in upper convex slope positions has
already resulted in yield losses of 30-40% of the average field yield. The
productivity losses are even more severe in years of climatic stress. However,
since on average these soils had a higher yield in no-till than in any other
tillage system, there seems little reason for farmers not to adopt the no-till
system immediately. The main mechanism of soil loss is tillage translocation,
and water erosion is likely to be minimal considering the high infiltration
capacities of sandy soils.
Conservation tillage systems appear to be increasing the soil's buffering
capacity against adverse climatic conditions, particularly drought stress.
This is encouraging considering the possibility of increased frequencies
of drought stress due to global warming trends. The full extent of the remediating
capabilities of no-till is not very well understood and should be the subject
of future studies. The remediation may appear to be faster than expected
because the massive soil loss from tillage translocation has been stopped
in the no-till, but continued in the conventional tillage system. Thus,
even if the yield productivity of severely degraded areas is stabilized,
it would appear that they have undergone an increase because of the continued
decline in productivity of the conventional system. The actual rate of remediation,
and degree to which it is possible to remediate these areas is unknown,
and beyond the scope of this study.
A considerable amount of research activity is currently underway regarding
variable management. A symposium on this subject will be held at the annual
American Society of Agronomy meeting. Other symposia have been held the
past few years by the American Society of Agricultural Engineers (ASAE).
Most of the activity has centred on development of spatial location sensors,
and variable application technology. There seems to be little doubt that
the technical ability to vary management within the field will be present;
the important missing link is the knowledge base to decide what kind of
management should be done at each point in the field. The yield data indicates
that reasonably consistent pattern of yield are present year after year.
Thus, variable management from a cost benefit point of view is necessary.
Technology for automatic measurement and recording of yield patterns is
available, but more accurate equipment is needed.
The consistent yield differences in the benchmarks from year to year
is encouraging, and it is clear that many of the differences are related
to soil loss or the effects of soil loss. The sites with very large yield
losses are diverging shoulder slope/crest positions as indicated by Battiston
et al. (1987). This problem can be attributed almost entirely to tillage
translocation. The obvious solution is not to till the fields. However,
it may be possible to change the pattern of tillage. The necessity of at
least alternating the direction of all tillage, including secondary tillage
cannot be overstated. The net soil losses from alternating tillage directions
is still very large, but much smaller than the loss from a single direction.
Without an accurate method of predicting the soil redistribution from
tillage translocation, there does not appear to be much hope in explaining
any more than about 40-50% of the spatial variations of soil properties.
The same constraint exists for predicting variations in yield response using
only landform classification. The alternative is to measure and map yield
response directly from year to year and identify sensitive areas from changes
in yield from year to year especially during climatic stress conditions.
While it is possible to predict that the areas of high yield loss will likely
be the shoulder slope positions, it is not true that all shoulder slope
positions have been eroded and thus will have yield reductions.
(From Technology Transfer Report Summaries - A. Hayes, L. Cruickshank,