A couple of months ago or so I got involved in a discussion on twitter about the effect of ten inch vs twelve inch row spacing on cereal crop yields and decided to make it the topic of a blog post. No one should argue the fact that row spacing which is *too* wide will result in inefficient utilization of space, sunlight, etc. within a field and yields would be less than optimal. Wider rows leave crops more vulnerable to losses from weed competition as well. However if the seeding operation was only about maximizing the utilization of the seed bed space, then we would just broadcast seed on the surface then incorporate and be done with it. The operation of seeding has other necessary but somewhat conflicting functions which favours seeding in rows (therefore limiting the utilization of space), such as ensuring uniform depth, good seed to soil contact, and applying fertilizer in way which allows the emerging crop timely access to nutrients.
The movement to direct seeding over the last 10 to 20 years has further increased and complicated the conflicting functions demanded of seeding equipment – now having to deal with getting through heavy surface crop residues, attaining precise seed depth and seed / soil contact, and delivering higher and higher rates of fertilizer in one pass operations. Wider row spacing is a design option for systems which helps deal with these increased challenges. But with the competing problems of optimum seed bed utilization on one hand, and proper seeding and fertilizer placement on the other, the inevitable question becomes: ‘What is the proper balance between optimum seed bed utilization (row spacing) vs crop residue management, quality seeding, and fertilizer placement?’
Over 13 seasons of crop consulting and agronomy I have worked with several different farmers using different (direct seeding) equipment and row spacing (up to 12 inches) in everything from drought to excessive moisture. Up to my twitter discussion a few months ago my view from these experiences had been that row spacing can interact with the unique conditions of each growing season in a manner which favours wider row spacing in one growing season and narrow row spacing in the next but that such differences tend to be offsetting when considered over the wide range of conditions over wider time spans. In academic literature, I viewed Lafond 1994 (L94) as seminal on the subject for the semi-arid Canadian prairies and consistent with my own experiences. However after the discussion I decided that spending some time revisiting the topic and reviewing some literature would be a worthwhile exercise and a good topic for a blog post.
I like L94 for many reasons. It is widely cited and to my knowledge the conclusions still stand as a strong case that wheat yield potential can be maintained with twelve inch row spacing in direct seeding systems. Direct seeding was an important distinction by Lafond in this and other work. According to Lafond, previous work on row spacing was set against more conventional tillage systems with little surface crop residue. With this distinction Lafond implies that inefficiency of resource utilization commonly attributed to wider rows might be countered by the gains in moisture use efficiency under direct seeding systems. The implication for row spacing is not that row spacing does not matter, but that perhaps the width ‘threshold’ – between where it doesn’t matter and does matter – is wider under a direct seeding system. Lafond also collected data on the recognized yield components of cereal crops as well: plant density, spike density, kernels per spike, and test weight to study the row spacing effects on these specific components. Variable Seeding rates and nitrogen rates were also evaluated separately and, along with row spacing, to check for interactions among these variables.
Guy Lafond, who sadly passed away this past April, published on this topic again, and most recently as lead author along with co-authors Bill May and Chris Holzapfel (LMH13) with similar conclusions studying oats. A key methodological difference in LMH13 is N placement. In L94 ammonium nitrate was surface broadcasted. In LMH13 N as urea was applied in the seeding operation via side band.
After reading through several papers and other publications on the effect of row spacing in cereal crop yields, I’m focusing on one (referenced by LMH13) as a comparison to L94. The paper is Chen, Neill, Wichman, and Westcott 2008 (C08) based on experiments in central Montana. I chose this paper because it is fairly current, geographically relatively close to Southern Saskatchewan, and allows for direct comparison of HRS wheat yield and yield components. L94 looked at 10, 20 and 30 centimeter row spacing while C08 used 15 and 30 centimeter row spacing. C08 reported a different conclusion then Lafond on the effect of row spacing on cereal crop yields. Like L94, C08 also reports on yield component data as well.
In reading the literature it seems widely recognized (and confirmed by both of these studies) that as row spacing increases, the plant density and spike density decreases. Lower plant density is attributed to greater concentration of seedlings per row in wider row spacing causing greater inter-seedling competition and therefore greater mortality (Amjad and Anderson, 2006). However, the other yield components can respond in a compensating and offsetting manner such as producing greater kernels per spike (Chen et al, 2008). Therefore it seems that a key questions is: Can lost yield potential in wider row spacing due to lower plant and spike density be recovered through increased kernels per spike and/or increased seed weight? Can direct seeding perhaps contribute to such a recovery?
L94 and C08 both looked at HRS wheat and both reported yield components as well as yield. Therefore, I can plot findings of both studies onto common graphs for a visual comparison. Below are two graphs showing crop density (figure 1) and spike density (figure 2). L94 is shown by the black trace and C08 in red. The y axis (vertical) shows units of density per square meter and the x axis (horizontal) shows row width in centimeters. These two figures confirm what has been seen widely in research into varying row width and mentioned above – that plant and spike density decreases with increasing row width. In L94, spike density decreases by 8% when moving from 20 to 30 cm and is consistent with C08 decreasing by 12% moving from 15 to 30 centimeters.
Next to look at is kernels per spike (figure 3) and the findings of these two studies here are in conflict with each other. L94 (black) shows increasing kernels per spike with increasing row width while C08 shows a very slight decrease with increasing row width.
Both papers show little change in seed weight over the differing row widths (Figure 4).
Figure 5 shows how the yields compared between the two experiments.
So it comes down to the differences in *one* yield component to account for the conflicting conclusions of L94 and C08 on row width and yield in HRS wheat. By expressing the data on a proportional scale, Figure 6 shows the almost perfect inverse relationship in L94 between spike density and kernels per spike with increasing row width. C08 on the other hand shows no such relationship between the components (Figure 7), with even a slight tendency for kernels per spike to vary in the same, rather than the opposite, direction as spike density. What could account for these conflicting results?
Going back to the introductory discussion in L94, Lafond does not dispute that previous work has found the tendency for yields to decrease as row spacing increases. However he does imply that direct seeding systems perform better by enabling more efficient utilization of moisture – in other words directly opposing (in part) previous explanations offered for declining yields with increased row spacing under conventional tillage. In C08 the Manning wheat was described as being planted into a “sweep tilled” annual cropping field therefore is a key difference in the two experiments as suggested by Lafond.
There was another interesting difference between the two methods which stuck out when I compared them: L94 applied the bulk of the nitrogen requirement by broadcasting ammonium nitrate on the surface. C08 states that liquid N “was dripped between wheat rows using a modified herbicide sprayer with special calibrated openers with attached drop tubes”. I read this to mean dribble band placement of N between rows. Dribble banded N would be progressively further from the seed row with widening row widths. Since efficiency of nutrient uptake (along with moisture and sunlight) is cited as accounting for lower yields with wider rows, one could at least question whether there is an (unaccounted for) effect in C08 of increasing distances from a nitrogen band that might confound with the row spacing effects.
Discussion and critiques of the post and arguments on the topic of row width in cereal production are welcome. I am interested particularly in discussing literature / data which might contribute to this discussion.
M. Amjad , and W. K. Anderson. 2006. Managing yield reductions from wide row spacing in wheat. Australian Journal of Experimental Agriculture 46: 1313–1321.
C. Chen, K. Neill, D. Wichman, and M. Westcott. 2008. Hard red spring wheat response to row spacing, seeding rate, and nitrogen. Agron. J. 100: 1296 – 1302.
T. Fiez, and W. Shillinger. 1999. Row spacing effects on direct seeded spring wheat and barley in the low-rainfall cropping zone. NW Direct Seed Cropping Systems Conference Procedings.
G. Lafond. 1994. Effects of row spacing, seeding rate and nitrogen on yield of barley and wheat under zero-till management. Can. J. Plant Sci. 74: 703 – 711.
G. Lafond, B. May, and C. Holzapfel. 2013. Row spacing and nitrogen fertilizer effect on no-till oat production. Agron. J. 105: 1-10.