Ulofe Uduokhai
University of Lethbridge
25/9/26
The cost of hybrid canola seed in Western Canada rose from about $2 kg⁻¹ for open-pollinated varieties in 1990 to about $24 kg⁻¹ by 2017, and growers have responded by reducing seeding rates, so that stands below 43 plants m⁻² are now common on farm fields. The reduction rests on two arguments: canola compensates for thin stands by branching, and the yield–density response is asymptotic, so additional plants return little yield for their cost. This synthesis asks whether the two arguments justify the densities now found on farms, and finds that neither does: compensation is effective only when environmental stress is low, and the economic optimum lies above the densities that many fields achieve. Field emergence also varies too widely for a seeding rate to deliver a predictable stand. Seeding rate is better read as a hedge against establishment failure and seasonal stress than as a yield optimum. On this reading the evidence favours seeding near 120 seeds m⁻² (12 seeds ft⁻²) rather than either the lowest or the highest rates tested, and identifies the prediction of establishment at seeding as the principal unresolved question.
Canola (Brassica napus L.) is a major crop of the Canadian Prairies, where its seeded area now rivals or exceeds that of wheat. The expansion is attributed largely to herbicide-resistant hybrids and their higher relative profitability, but these technologies raised the cost of seed. Open-pollinated varieties cost approximately $2 kg⁻¹ in 1990; hybrid seed cost approximately $24 kg⁻¹ by 2017, excluding technology use fees, while the price of canola rose only from about $300 to $560 t⁻¹ over the same period (Hartman & Jeffrey, 2021). The ratio of seed cost to crop price has therefore changed markedly, and with it the economics of the seeding rate decision.
Growers responded to the cost of hybrid seed by seeding less of it. Alberta producers in the early 1990s mostly seeded 5.6 to 7.8 kg ha⁻¹; recent grower surveys report 4.5 to 5.5 kg ha⁻¹, and because hybrid seed is larger, fewer seeds are placed per unit area than the change in mass alone suggests (Hartman & Jeffrey, 2021). The practice is defended on two grounds. Canola plants in thin stands produce more branches and pods, which is taken to compensate for the missing plants. And the yield–density relationship is non-linear and asymptotic, so that each additional plant adds less yield than the one before (Dhillon et al., 2022). On this basis a low seeding rate is treated as a sound economy rather than a yield risk.
The consequence is visible in surveys. In the 2017 Alberta Weed Survey, 60% of the canola fields measured had fewer than 43 plants m⁻², below the current recommendation of 50 to 80 plants m⁻² (Hartman & Jeffrey, 2021; Canola Council of Canada, n.d.). The remainder of this synthesis asks whether the compensation and asymptote arguments can support densities this low, and whether a seeding rate can be relied on to produce a chosen density at all.
The ability of canola to compensate for thin stands is real but conditional. In a multi-site study conducted between 2018 and 2022 at irrigated and rain-fed sites near Lethbridge and at Lacombe, Indian Head, and Brandon, seed number and seed weight on secondary branches became the critical yield components at low seeding densities, but only when environmental stress was low (Beres et al., 2025). Irrespective of hybrid or harvest method, 120 and 180 seeds m⁻² gave higher and more stable yields than 60 seeds m⁻². Compensation also carries a cost of its own: plants in the thinnest stands developed more branches and pods but lagged in development by up to five days, and the resulting spread in maturity is associated with seed loss and poorer harvestability (Beres et al., 2025).
Branching therefore does not guarantee that a thin stand will recover its yield. It is a response that operates in favourable seasons, and in Western Canada the season is not known at the time of seeding. The argument from compensation holds for the low-stress case and is silent on the drought and pest pressure that define many Prairie seasons.
The asymptotic form of the yield–density relationship is well supported. Across 12 site-years in southern Alberta, plant density rose with seeding rate along a negative exponential curve, and yield approached a plateau with increasing density (Dhillon et al., 2022). The same study found that the relationship was only weak to moderate in strength at most site-years, which is to say that density explained a limited part of the variation in yield from one site-year to the next.
An asymptote, however, justifies reducing density from the level that maximises yield, not to any level at all. A meta-analysis of 85 unique site-years of Western Canadian hybrid canola data estimated the economically optimal density at 62 to 73 plants m⁻², assuming 5 g thousand-seed weight, 60% emergence, and seed and crop prices of $26 kg⁻¹ and $0.45 kg⁻¹ (Hartman & Jeffrey, 2021). This range already accounts for the cost of hybrid seed and is consistent with industry recommendations, yet it lies well above the fewer than 43 plants m⁻² measured in most surveyed fields. The authors concluded that many growers may have over-compensated for seed cost by reducing seeding rates too far. The asymptote argument is therefore correct in form but has been applied beyond the range in which it holds.
Both arguments assume that a grower who chooses a seeding rate thereby chooses a plant density. The emergence data do not support this. A meta-analysis of 47 site-years of small-plot trials in Saskatchewan and Manitoba between 2013 and 2022 found a mean field emergence of 60.7%, but values ranged from as low as 20 to 30% to as high as 80 to 90% (Catellier, 2023). Seeding date and air temperature before and after seeding were the most influential variables, probably through their effect on soil moisture, and emergence also declined as seeding density increased.
The field trials show the same gap between seed and plant. In the Prairie-wide study, 120 and 180 seeds m⁻² established on average 57 and 80 plants m⁻², under half of the seed sown (Beres et al., 2025). Seeding equipment adds a further source of variation: narrow-row precision planting increased yield by an average of 463 kg ha⁻¹ (10%) over an air drill under irrigation, whereas under water-limited conditions the air drill yielded as much or more, and a wide-row planter gave poor establishment in both systems because of high in-row plant density (Dhillon et al., 2022).
A seeding rate calculated from average emergence will therefore fall short of its target in a substantial fraction of fields, and the shortfall cannot be known until after emergence. A low target density leaves no margin for this error, whereas a moderate one absorbs it.
If low rates carry yield risk, it does not follow that the highest rates are most profitable. In the Prairie-wide study, the late-maturing hybrid L255PC yielded 2.66 t ha⁻¹ (47 bu ac⁻¹) when sown at 60 seeds m⁻² and 2.86 t ha⁻¹ (51 bu ac⁻¹) at 120 or 180 seeds m⁻², with the highest and most stable yields under straight cutting (Beres et al., 2025). At 2023 canola prices, the net return from 180 seeds m⁻² was $196 ha⁻¹ ($79 ac⁻¹) lower than from 120 seeds m⁻² and $123 ha⁻¹ ($50 ac⁻¹) lower than from 60 seeds m⁻²; the intermediate rate returned about $72 ha⁻¹ ($29 ac⁻¹) more than the lowest (Canadian Agronomist, 2025).
The return is therefore not monotonic in seeding rate. The highest rate added no yield over the intermediate rate and was the least profitable of the three, while the lowest rate lost both yield and stability. The case against low seeding rates is not that more seed always pays, but that a moderate rate buys stability that the lowest rate lacks, at a cost the yield recovers. The location of this optimum is sensitive to seed cost, crop price, seed size, and emergence (Hartman & Jeffrey, 2021), so the figure of 120 seeds m⁻² is a present estimate, not a constant.
The evidence reviewed here locates the main uncertainty in establishment rather than in the yield response itself. Hartman and Jeffrey (2021) noted that the yield response to density varied widely and could not be predicted at seeding, and called for decision tools to support planting decisions. The emergence meta-analysis suggests that part of this variation is predictable from seeding date, temperature, and moisture (Catellier, 2023), but to the author's knowledge no model yet converts these variables into a field-specific seeding rate at the time of seeding.
Three questions follow. First, whether emergence can be predicted from conditions known at seeding with enough precision to adjust seeding rate field by field. Second, how precision planting interacts with seeding rate across the moisture gradient, since its advantage appears under irrigation and disappears under water limitation (Dhillon et al., 2022), and whether a more uniform stand would make lower rates safer (Beres et al., 2025). Third, whether thin stands carry costs beyond the current season, since lower densities have been associated with poorer weed control and greater selection for herbicide-resistant weeds (Beres et al., 2025). The paired irrigated and rain-fed sites near Lethbridge provide a natural contrast for the first two: if stress governs the density response, the benefit of higher rates should be small under irrigation and large under dryland conditions within the same seasons.
The reduction of seeding rates in hybrid canola rests on arguments that are correct in principle and overextended in practice. Compensation by branching recovers yield when stress is low, not across the range of Prairie seasons, and the asymptotic yield response supports an economic optimum of 62 to 73 plants m⁻², above the densities now common on farms. Because emergence ranges from roughly 20 to 90%, the seeding rate does not fix the stand, and a low target leaves no margin for establishment failure. Seeding rate is better understood as insurance against establishment and seasonal risk, priced by the cost of seed. On current evidence and prices the balance lies near 120 seeds m⁻², and the most useful next step is not a further estimate of the yield–density curve but a means of predicting establishment at the time of seeding.
Beres, B. L., Wang, Z., Stevenson, F. C., Geddes, C. M., Tidemann, B. D., Kubota, H., May, W. E., & Mohr, R. M. (2025). Optimizing canola production in the Northern Great Plains by leveraging genotype × environment × management synergies. Crop Science, 65, e70115. https://doi.org/10.1002/csc2.70115
Canadian Agronomist. (2025, October 15). Optimizing canola production. https://canadianagronomist.ca/optimizing-canola-production/
Canola Council of Canada. (n.d.). Target 5 to 8 plants per square foot. Canola Watch. https://www.canolacouncil.org/canola-watch/fundamentals/target-5-to-8-plants-per-square-foot/
Catellier, C. (2023). A meta-analysis of small-plot trial data to examine the relationship between crop development and environmental conditions in canola [Final report]. Indian Head Agricultural Research Foundation. https://iharf.ca/wp-content/uploads/2025/05/A-meta-analysis-of-small-plot-trial-data-to-examine-the-relationship-between-crop-development-and-environmental-conditions-in-canola.pdf
Dhillon, G. S., Baarda, L., Gretzinger, M., & Coles, K. (2022). Effect of precision planting and seeding rates on canola plant density and seed yield in southern Alberta. Canadian Journal of Plant Science, 102(3), 698-709. https://doi.org/10.1139/cjps-2020-0186
Hartman, M. D., & Jeffrey, S. R. (2021). Estimating the economic optimal target density of hybrid canola based on data from a western Canadian meta-analysis. Canadian Journal of Plant Science, 101(3), 393-407. https://doi.org/10.1139/cjps-2020-0162