Direct-seeding systems are the most rapidly evolving agricultural technology in western Canada. Economic and soil conservation benefits have been documented at the farm and research level and have become the driving forces behind producer adoption of these conservation-tillage systems. Because the term direct seeding encompasses a broad range of one- and two-pass fertilization and seeding systems, questions have arisen regarding the relative efficiency of different approaches. The most commonly asked questions relate to the effects of soil disturbance and fertilizer placement on crop yields, weed management, and production economics.

Field research was initiated at Brandon MB, Melfort SK, and Beaverlodge AB to determine the impact of the different timings and placements of nitrogen (N) on wheat and canola agronomy and weed dynamics.

Materials and Methods

Studies were established in 1996 on AAFC research stations at the respective locations. The experiments were set up in a split-split plot design with 4 replications. Plots were 7.3m (24') X 15.0m (50') in size. Crops (wheat and canola) grown in rotation were the main plot, fertilizer treatments (placement and timing) were the sub-plots, and herbicide rate used was the sub-sub plot. For the sake of statistical analysis, all agronomic data were analysed separately by crop. In order to reduce site to site variability, seeding was done at all sites using identical air-seeders setup to seed on 22.5 cm (9") and 30 cm (12") row spacings (one pair of seeders at each site). With these seeders MAP fertilizer (11-51-0) was seed placed and for side banding treatments urea nitrogen fertilizer (46-0-0) was side banded 2.5cm (1") to the side and 2.5cm (1") below the crop row. The high disturbance seeding treatments (sweep seeding) were seeded with 30 cm (12" inch) shovels on 22.5 cm (9") spacing (using the 22.5 cm row spacing seeder). To ensure uniformity, sweeps were purchased from a common source and used at all locations.

Table 1. Treatments used to compare one and two pass, and high and low disturbance seeding systems at each site.

Each of these fertilizer treatments was present in wheat and canola with the two crops rotating on an annual basis. To determine the long-term impact of adopting one of these approaches, the fertilizer treatments remained constant from year to year (eg: fall banding treatments always occurred on previous fall banding plots). The experiments were established to run for five years. To determine if efficient fertilizer usage by the crops would reduce weed competition, each of the fertilizer treatments was duplicated at 100% and 66% of recommended rates of Horizon plus Target and Roundup in wheat and canola, respectively (herbicide plots remained constant from year to year eg. the 100% rate was used every year in the same plot regardless of crop). This resulted in 10 fertilizer/herbicide treatments in each crop per replicate for a total of 80 plots per site.

Agronomic details: To best determine the efficiency of nitrogen fertilizer, it was applied at 66% of typical recommended rates. This was 85kg/ha at Melfort, 70 kg/ha at Beaverlodge, 65 kg/ha at Brandon with N rates adjusted for nitrogen in MAP which was seed placed at soil test recommendation levels. Sulphur fertilizer was applied as needed at each site for canola production. Teal wheat was seeded at 160 kg/ha. Quest canola (Round-Up/glyphosate tolerant) was seed at 7 kg/ha plus granular insecticide for flea beetle control. Pre-seeding, pre-harvest, and post-harvest herbicide treatments were applied as required at each site. In-crop use was kept common at all sites with Roundup (glyphosate) applied to canola at a rate of 1.24L/ha for the 100% rate treatment and Horizon (clodinafop propargyl) plus target in wheat at 230ml/ha for the 100% rate. Target (MCPA + mecoprop+ dicamba) was applied as a tank mix with Horizon to control most broadleaf weeds at all sites with the 100% rate being 1.0 L/ha. Fungicides were applied as needed for plant diseases at each site.

Crop stand counts, head counts, height, Haun stage (wheat only) were evaluated to compare the impact of the treatments on crop growth. Whole plots were harvested and canola was swathed prior to machine harvesting. Grain samples were cleaned to grain commission standards and corrected for differences in moisture prior to data analysis. To compare nutrient dynamics, weed and crop biomass samples were collected at crop heading and analysed for nutrient content. Protein levels were determined for the grain samples. Soil samples were taken in the spring and fall of each year from each plot to determine moisture usage and in the fall for nutrient analysis. Weeds were identified by species and counted in 20 0.5 X 0.5 m² (0.25m²) quadrats per plot prior to seeding, before in-crop spraying and in late July each year. Since this the purpose of this paper was to summarize the trends during the first 3 years of the trials, only yield data and weed counts will be presented.

All data were subjected to an analysis of variance with means separated by determining the Least Significant Difference (p<0.05) or orthogonal contrasts (p<0.05). For the sake of data summarisation the number of site years where statistical differences occurred, and the rank of least weedy and rank of highest yields for each treatment were presented. Sixty seven weed species were present within the study including typical annual species, perennial species, and volunteer crops as weeds with about 30 species present in a site year. Total weed density information was used to summarize the weed data for this paper.

Results and Discussion

When statistical differences in crop yields were compared for the 9 site years of data (1996, 1997, 1998 at each of the three sites), no one fertilizer treatment was superior in the majority of cases and in all but one case reducing the rate of herbicide in wheat or canola did not reduce yield (Table 2). Fall banding N was better than spring banding at Melfort for wheat in 2 of 3 years while spring banding was better in 1 of 3 years at each site. Seeding wheat on the narrow row spacing was superior to the wide row spacing for 2 of 3 years at Brandon and Melfort and 1 of 3 years at Beaverlodge. Seeding canola on the narrow row spacing was superior at Brandon in 2 of 3 years, but did not have the same effect at Melfort or Beaverlodge. In no case was the wide row spacing superior to the narrow for wheat or canola yield. Seeding with sweeps had a positive impact on wheat yield in some cases and a negative impact on others while seeding canola with sweeps had a negative impact in 2 of the 9 site years.

To further compare differences between treatments data were ranked from highest to lowest yields (Table 3). Since herbicide rates were not significantly different, only the 5 fertilizer treatments were ranked. Again the data indicated that no one treatment was the best over all, but that the most consistent treatment was the 20 cm (9") side banded treatment for both crops, especially at Brandon and Melfort. However, considerable variability existed with any one treatment being ranked from highest to lowest yield depending on site and year and crop. For example, at Brandon the sweeps were generally ranked 4th or 5th while the reverse was true at Melfort. Furthermore, while sweeps were ranked 1st for wheat in 1996 and 1997 they ranked 4th and 5th for canola grown during the same years at Beaverlodge. Therefore, the pursuit of the ultimate seeding system may be less important than other aspects of crop agronomy in direct-seeding systems.

When the impact of fertilizer and herbicide management on weeds was compared for statistical differences over the 8 site years of data (data not available for Melfort in 1996) no one treatment was consistently the most or least weedy (Table 4). Prior to seeding, fall banding treatments in wheat (canola stubble) were weedier than other treatments. For canola, fall banding was weedier in 2 of 8 site years, but spring banding was weedier in 3 of 8 site years. Particular attention should be paid to the pre-spray weed counts since this is when most of crop yield losses due to weeds occurs. At this time spring banding fertilizer was weedier than fall banding in 4 of 8 site years in wheat (canola stubble), but fall and spring banding each were weedier than the other in 2 of 8 situation for canola (wheat stubble). When differences occurred, random banding (fall or spring) was weedier than side banding. Sweeps were more often weedier at the in-crop stage. These data suggest that side banding has the greatest potential to reduce crop losses due to weeds. When statistical differences occurred at the July time frame the residual weeds were generally greater in fall versus spring banded treatments, wide versus narrow row spacings, and in sweep treatments. In only 1 of 8 site years was the reduced herbicide rate plots weedier than high rate plots at the in-crop time frame. This indicates that potential crop yield losses due to increased weediness in reduce rate treatments may not be as great as has been thought; however, in 7 of 8 cases reduced rate plots were weedier in late July indicating the potential for increased losses due to weed pressure does exist.

When treatments were compared by ranking weediness, the fall banded and sweep seeded treatments tended to be the weediest while the side banded treatments tended to be the least weedy prior to seeding the crop (Table 5). At the time of in-crop spraying, when crop losses due to weed competition occurs, fall banded wheat plots were the least weedy followed by side banded treatments with sweeps and spring banding tending to be the most weedy (Table 6). A consistent pattern of increased or decreased weediness due to low rates of herbicide being used was not evident at either the pre-seeding or the in-crop time of assessment. Conversely, in July weediness ranked higher for plots receiving the low rate of herbicide (Table 7). At this time the wide row spacing treatments ranked as weedier compared to those with narrow row spacings. Due to variability between sites, years, and crops it was difficult to find consistent differences among the five fertilizer management systems based on ranking weediness in July.

Conclusions

The row spacing comparisons require further discussion. The lack of increased weediness of the wide row spacing treatments and the increased weediness of the narrow row spacing treatment (1 of 8 site years only) at the time of in-crop spraying indicates that the potential for crop yield loss due to weed competition may not neccesarily be greater at the wide row spacing. However, there was an increased weediness of the wide row spacing in July and lower crop yields in this treatment. Past research showing wide row spacings do not have lower yields was conducted without side banding N. The urea fertilizer being placed 2.5 by 2.5 cm (1" by 1") to the side and below the seed may have caused the yield reduction and subsequent increased weediness of the wide row spacing treatment due to the higher concentration of N in the band (same rate per unit area in narrow and wide row spacings). Furthermore, this problem for side banding may be alleviated for any row spacing by changing the form of N to a potentially less toxic source, using urease inhibitors to delay the release of N from urea, and physically increasing the separation between seed and fertilizer. The main point of these results is that there is greater potential for crop injury from side banding urea than was previously thought, but that side banding over all performed well from a weed point of view compared to random banding or high disturbance sweep seeding and therefore requires further research to get the full potential benefits from one pass seeding.

Caution should be used when interpreting the data presented in this paper because using ranked rather than absolute values may give the impression that larger or smaller differences occurred. For example, yields among the fall, spring, narrow row spacing, and sweep treatments in wheat were often very close despite differences in rank. Conversely, although the narrow and wide row spacing canola treatments were often ranked close together, large differences in yield occurred in some instances.

Several generalizations can be made from the data after three years of research, but the full story will become evident after 5 years. In general, the low rate herbicide treatments have not led to lower yields despite weedier conditions in July of each year. This may be due to lack of differences in weed pressure at the time of in-crop spraying suggesting that the occasional use of lower than registered herbicide rates will not necessarily lead to production problems. The fertilizer/disturbance treatment that has been most consistent for yield and weediness to date for wheat and canola has been the narrow row spacing side banded treatment. However, it is worth noting that no one treatment performed the best every year or at every location. Future data will help to elucidate interactions between herbicide rates and fertilizer/disturbance treatments.

Acknowledgements

The authors acknowledge the primary support and vision for multi-year and multi-discipline research from the Canadian Fertilizer Institute and Agriculture and Agri-Food Canada’s MII program as well as the support of Norvartis Canada Inc. and Monsanto Canada Inc which has made this research project possible.