About Us | Help Videos | Contact Us | Subscriptions

Agronomy Journal - Article



This article in AJ

  1. Vol. 104 No. 4, p. 1023-1031
    Received: Dec 19, 2011

    * Corresponding author(s): william.may@agr.gc.ca


The Effects of Chloride and Potassium Nutrition on Seed Yield of Annual Canarygrass

  1. William E. May *a,
  2. Sukhdev S. Malhib,
  3. Christopher B. Holzapfelc,
  4. Bryan X. Nybod,
  5. Jeff J. Schoenaue and
  6. Guy P. Lafonda
  1. a Agriculture and Agri-Food Canada, Indian Head Research Farm, RR#1 Gov Rd., P.O. Box 760, Indian Head, SK, Canada S0G 2K0
    b Agriculture and Agri-Food Canada, Melfort Research Farm, P.O. Box 1240, Melfort, SK, Canada S0E 1A0
    c Indian Head Agricultural Research Foundation, RR#1 Gov Rd, Box 156, Indian Head, SK, Canada S0G 2K0
    d Wheatland Conservation Area Inc., P.O. Box 2015, Swift Current, SK, Canada S9H 4M7
    e Univ. of Saskatchewan, Dep. of Plant Sciences, 51 Campus Dr., Saskatoon, SK, Canada S7N 5A8. Received 19 Dec. 2011


The year-to-year variability of seed yield in annual canarygrass (Phalaris canariensis L.) is a major concern among growers. A field experiment was conducted at 13 site-years across Saskatchewan to determine the response of annual canarygrass seed yield to K and Cl, and to provide better recommendations to producers on the use of KCl fertilizer in annual canarygrass based on soil test results. Potassium did not affect the yield or development of annual canarygrass over a range of 155 to 717 kg K ha−1 in the top 15 cm of soil. Chloride had a large impact on annual canarygrass seed yield; seed yield increased by approximately 24% when Cl was added in the form of KCl or CaCl2 when averaged across all sites. The seed yield increased because the application of Cl increased panicle size (seeds panicle−1). The magnitude of the response tended to increase as level of Cl in the soil decreased. Annual canarygrass growers need to measure Cl when using soil tests to determine fertilizer requirements. It is recommended that 9.1 kg Cl ha−1 in the form of 20 kg ha−1 of KCl be applied when the Cl level in the surface soil (0–15 cm) is below 70 kg Cl ha−1. The findings encourage growers to conduct individual field test strips to determine the strength of the Cl response.


    LV, latent variables; NDVI, normalized difference vegetative index; PLS, partial least squares projection to latent structures; VIP, variable importance in the projection

Historically, the greatest use for annual canarygrass seed has been as bird food. Approximately 69 to 79% of the world's annual canarygrass is produced in Canada (FAO, 2008). Production of annual canarygrass in North America is centered in the province of Saskatchewan (FAO, 2008). Annual canarygrass was first tested as a grain crop in 1906 at Indian Head, SK, (MacKay, 1907). In Canada, recording the area seeded to annual canarygrass began in 1971 with 800 ha and the seeded acreage has ranged from 95,000 to 350,000 ha over the last 20 yr, with 89 to 98% of this production occurring in Saskatchewan (Saskatchewan Ministry of Agriculture, 2008).

Annual canarygrass growers identified spatial and temporal variability in seed yield as their greatest concern in an informal survey (May, 1998). The survey was conducted by separately interviewing between 26 individual farmers at their homes or after meetings. The most common comment from growers was the large amount of biomass being produced in certain years with very little seed being harvested. Research on various agronomic practices which include seeding date, seeding rate, insect, disease and weed control and fertility on annual canarygrass seed yield has been conducted over the years (Holt and Hunter, 1987; Holt, 1988 , 1989; Miller, 2000; May, 2002; May et al., 2012). May et al. (2012) and Miller (2000) found that seeding date affected seed yield and could account for some but not all of the seed yield variability reported by growers. Seeding rate has also been examined and there was a modest increase in seed yield of 12.5% in one study (Holt, 1989), and 8.5% in a second study (May et al., 2012). In the second study, the increase in seed yield was combined with increased yield stability as seeding rate increased (May et al., 2012). Weed control is important in preventing a large decrease in seed yield (Holt and Hunter, 1987); however, weed control issues did not appear to be a major cause of the observed variation in seed yield reported by farmers. Disease control is important in annual canarygrass and appears to account for some of this seed yield variability (May, 2002).

The effects of N fertility on annual canarygrass have been studied. Holt (1988) found over a 6-yr period at Indian Head, SK, that seed yield was maximized between 50 and 75 kg N ha−1, increasing yield by 43% to 1680 kg ha−1. This study was conducted using a conventional tillage system and N rates between 0 and 100 kg N ha−1. May et al. (2012) in a study conducted across Saskatchewan, found that grain yield was maximized at 78 kg N ha−1 increasing seed yield by 7% to 1216 kg ha−1. This study was conducted in a no-till system evaluating N rates ranging from 20 to 100 kg N ha−1. Nitrogen does not have a large enough effect to account for the reported yield variation (May, 1998).

Very little information has been published on the requirements of annual canarygrass for K and Cl. Putnam et al. (1996) published K recommendations for annual canarygrass in North Dakota. In a study examining the effects of K and S, KCl had a large effect on seed yield at specific locations while S had no effect on seed yield of annual canarygrass (May, 2002). The response could not be correlated with K levels in the soil, and Cl levels in the soil were not tested.

Potassium is an osmotically active element associated with cell expansion, enzymes activation, and regulation of stomata (Amtmann et al., 2008; Romheld and Kirkby, 2010). Potassium is also involved in photosynthesis, phloem loading, uptake and transport, and the storage of assimilates. Potassium deficiency can be manifested in many ways including, slow growth, lower yield, lodging, increased susceptibility to fungal and bacterial diseases and increased sensitivity to drought (Beaton and Sekhon, 1985; Amtmann et al., 2008; Romheld and Kirkby, 2010).

The effect of Cl has been not studied in annual canarygrass; however, Cl is involved in several plant functions, including, enzyme activation, photosynthesis, nutrient transport, stomatal activity, accelerated plant development, and improved disease suppression (Fixen,1993). The effect of Cl has been studied in wheat (Triticum aestivum L.), barley (Hordeum vulgare L.), and oat (Avena sativa L.), but the results have varied between studies. Mohr et al. (1995a , 1995b) found an inconsistent response in the grain yield in wheat and barley to Cl. Fixen (1993) found that the grain yield of spring wheat responded to Cl and published a chart on when to apply Cl to spring wheat based on soil residual Cl. In wheat and barley, cultivars have differed in their response to Cl (Mohr et al., 1995a , 1995b; Grant et al., 2001; Evans and Riedell, 2006). When researchers have found an effect of Cl they have attributed the response to several factors including, osmoregulatory functions, interactions with other nutrients, interaction with diseases and alteration of crop development (Fixen, 1993). Many of the studies could not identify the reason for the observed effect of Cl in increasing grain yield (Fixen et al., 1986; Engel et al., 1994; Gaspar et al., 1994; Diaz-Zorita et al., 2004).

The objectives of the study were to determine if K or Cl or both are responsible for any effects on seed yield of annual canarygrass and possibly help to explain the reported temporal variability in seed yield. Field experiments were conducted to determine the responsiveness of annual canarygrass on seed yield to K and/or Cl, and to provide better recommendations to producers on the use of KCl in annual canarygrass using soil test results.


A field experiment was conducted at four areas in Saskatchewan, Canada Indian Head (50°33’08.37’’ N, 103°38’39.82’’ W and elevation 579 m), Carry The Kettle (2 sites) (50°24’56.59’’ N, 103°34’47.13’’ W and elevation 642 m, 50°24’46.98’’ N, 103°35’39.29’’ W and elevation 616 m), Regina plain in 2007 (50°20’59.67’’ N, 104°47’20.50’’ W and elevation 576 m) in 2008 and 2009 (50°08’13.17’’ N, 104°21’41.51’’ W and elevation 580 m) and Stewart Valley (50°34’26.94’’ N, 107°49’30.99’’ W and elevation 722 m) from 2007 to 2009. The soil series were Indian Head heavy clay (Orthic Vertisol or Haplocryert) at Indian Head, an Oxbow loam (Orthic Black Chernozem or Udic Boroll) and Ellisboro (Rego Black Chernozem or Udic Boroll) at Carry The Kettle, Regina heavy clay (Orthic Vertisol or Haplocryert) on the Regina plain and Sceptre heavy clay (Orthic Vertisol or Haplocryert) at Stewart Valley. To differentiate between the two sites near Carry The Kettle when needed they will be called CTK ellisboro and CTK loam. The site at Stewart Valley in 2008 was not used due to a salinity and moisture gradient that was parallel to the replicates that created considerable variability among plots. The experiment was conducted at Indian Head only in 2008 and 2009.

The experiment design used for each site-year was a randomized complete block design involving 10 treatments and four replications. The treatments consisted of a check plot with no K or Cl and three forms of fertilizer, KCl, K2SO4, and CaCl2 each applied at three rates (tbl1Table 1). The three rates of fertilizer for KCl and K2SO4 resulted in 10, 20, and 30 kg K ha−1 being applied and three rates of fertilizer for KCl and CaCl2 resulted in 9.1, 18.2 and 27.3 kg Cl ha−1 being applied. This resulted in three treatments with 10, 20, and 30 kg K ha−1 from K2SO4, three treatments with 9.1, 18.2 and 27.3 kg Cl ha−1 from CaCl2 and three treatments with the three rates with K and Cl in combination from KCl. KCl is the form of fertilizer used by producers to apply K or Cl. A blanket application of fertilizers was applied to all plots at a rate of 46 kg N ha−1 [urea, CO(NH2)2], 10.9 kg P ha−1 (mono-ammonium phosphate, NH4H2PO4), 12 kg ha−1of S [ammonium sulfate, (NH4)2SO4]. All fertilizers were side banded and placed 2 to 3 cm to the side and 7 to 8 cm below the seed using a hoe opener.

View Full Table | Close Full ViewTable 1.

The form and rate of K and Cl used for each treatment.

Treatment Form Potassium Chloride KCl rate
K2SO4 rate
CaCl2 rate
0–0–60–0† 0–0–51–17† (94% pure CaCl2)
kg ha−1
1 none 0 0.0
2 KCl 10 9.1 20.0
3 KCl 20 18.2 40.0
4 KCl 30 27.3 60.0
5 K2SO4 10 23.5
6 K2SO4 20 47.1
7 K2SO4 30 70.6
8 CaCl2 9.1 15.1
9 CaCl2 18.2 30.2
10 CaCl2 27.3 45.3
The concentration of K in the applied fertilizer is calculated as K2O.

The cultivar CDC Togo was seeded at a rate of 35 kg ha−1 at all locations. This cultivar is a glabrous (hairless) type of annual canarygrass. The row width was 30.5 cm at Indian Head, Carry The Kettle and Regina, and 25.4 cm at Stewart Valley. The plot size was 10.7 by 4.0 m at Indian Head, Carry The Kettle and Regina; and 9.1 by 4.3 m at Stewart Valley. The plots were managed using a no-till production system. Glyphosate [N-(phosphonomethyl) glycine] was applied before seeding and all in-crop broadleaf herbicide applications were determined separately for each location according to weed species and density using recommended products and rates (Saskatchewan Ministry of Agriculture, 2010).

Data Collection

Soil tests were performed at each site for N, P, K, S, and Cl. Spring soil test levels of NO3– N and Cl were measured to a depth of 60 cm. This was done by collecting soil from a 0- to15-cm depth and a 15- to 60-cm depth. Soil residual P, K, and S were measured in a 0- to 15-cm depth. A NaHCO3 extraction procedure (Hamm et al., 1970) was used to estimate residual soil N (NO3), P, and K. Available Cl was determined by extraction of 5 g of soil with 50 mL of water followed by filtration and determination of Cl in the filtered extract using a Technicon Auto-analyzer II (Inland Waters Directorate, 1979). Available S was determined by extraction of 10 g of soil with 50 mL of 0.001 M CaCl2 followed by filtration and determination of S in the filtered extract using a Technicon Auto-analyzer II (Hamm et al., 1973).

The bioavailable K supply rates were determined using ion exchange resin membranes probes (Plant Root Simulator [PRS] probes, Western Ag Innovations, Saskatoon, SK, Canada) according to the procedures outlined in Qian and Schoenau (2002) and Qian et al. (2008) . This measurement of K availability provided an indication of the supply of readily available K to an adsorbing surface, taking into account replenishment of soil solution K by the solid phase as well as movement by diffusion. The probes were charged by soaking in 0.5 M HCl for 2 to 4 h to saturate the exchange sites with H+ ions. The cation probes were then inserted directly into a sample of soil at field capacity for 24 h. After 24 h, the probes were removed and washed of all soil particles and placed into a clean self-sealing bag and treated with 20 mL of 0.5 M HCl for 1 h to elute the sorbed K ions from the membrane surface. The eluent was the analyzed for K concentration using flame emission spectroscopy. Using the concentration of K in the eluent and the surface area of the membrane, K supply rate was calculated as μg K sorbed per cm2 of membrane surface over 24 h.

Plant density was determined 3 to 5 wk after seeding and annual canarygrass panicles were counted after panicle emergence. Both plants and panicles were measured in two 1-m sections of crop row within each plot. Physiological maturity was reached when kernel moisture was approximately 30 to 35%. Lodging was rated in each plot at physiological maturity using a scale of 1 to 10 (1 = standing, 10 = completely lodged). At each location, the mean normalized difference vegetative index (NDVI) from each plot was measured between stem elongation and flag leaf using a handheld optical sensor (RT 100, GreenSeeker, NTech Industries, Ukiah, CA) (NTech Industries, 2009). The NDVI is a vegetation index that is an indirect measure of the crops aboveground biomass and nutrient uptake (Moges et al., 2004; Freeman et al., 2007; Osborne, 2007). Significant correlations between NDVI and grain yield have been obtained in winter wheat (Raun et al., 2001), corn (Zea mays L.; Teal et al., 2006), soybean [Glycine max (L.) Merr.; Ma et al., 2001], spring wheat (Osborne, 2007), grain sorghum [Sorghum bicolor (L.) Moench subsp. bicolor; Moges et al., 2007; Tucker and Mengel, 2008] and canola (Brassica napus L.; Holzapfel et al., 2009). The GreenSeeker optical sensor determines NDVI by actively emitting radiation in the visible red (∼660nm) and near infrared (∼770nm) bandwidths and measuring the proportion of emitted radiation that is reflected from the canopy. The sensor's field of view is approximately 60-cm wide and NDVI is calculated using the following relationship, NDVI = (NIR – red)/(NIR + red), where red and NIR are the spectral reflectance measurements for the visible red and near-infrared regions, respectively. Depending on the plot length, approximately 30 to 70 individual NDVI values were logged for each plot and the average of these was used to represent the plot.

Seed yield was expressed on a clean seed basis and was corrected to a uniform kernel moisture content of 13%. Kernel weight, expressed per 1000 seeds (g), was calculated by weighing between 700 and 1000 kernels with the number of kernels determined using an automated seed counter. Kernel panicle−1 was calculated using panicles m−2, seed yield, and kernel weight. Kernel m−2 was calculated using seed yield and kernel weight. Test weight was measured as specified by the Canadian Grain Commission's Official Grain Grading Guide (2011).

Statistical Analysis

A combined analysis was conducted using the MIXED procedure in SAS (Littell et al., 2006). The effect of replicate and site (location × year combinations) were considered random, and the effects of applied treatments were considered fixed. A combination of variance estimates and P values were used to determine the importance of the random sites by fixed effects interaction. Treatment effects were declared significant at P ≤ 0.05.

We then examined important site × treatment interactions. The first portion of this analysis considered the quantitative relationships between crop yield (dependent variable) vs. site and environment indicators using the partial least squares projection to latent structures (PLS) method. Data for the PLS analysis consisted of a matrix with each site as a row (site means were used for crop dependent variables), and the site and environment indicators as columns. The PLS analysis was performed using the PROC PLS procedure of SAS (Tobias, 1995; SAS Institute, 2005). Initially, all site and environment indicators measured in the study were included as predictor variables in the PLS model. Predictors with the greatest influence in explaining dependent variable variability, or variable importance in the projection (VIP), were then selected based on the criterion of VIP > 0.8 (Wold, 1994). It was found that three PLS latent variables (LV) typically explained close to 80% or more of the variation for the predictor and response variables. The X loadings represent the correlation between raw data for each of the site and environment predictors to the scores for each LV. These X loadings were used to assess the relative importance of each of the site and environment predictors toward each of the three LVs. The LV scores were used for the second portion of the analysis to further examine important site × treatment interactions for response variables of interest.

The second portion of this analysis, which used LV scores and original data, was conducted to determine reasons for important site × crop/cultivar treatment interactions determined from the initial univariate mixed model analysis of variance. Scores for each LV were merged with the original data set including crop yield. This meant there was a single score available for each site × LV combination. An extension of the previously described mixed model was implemented to explore site interactions with treatments (Littell et al., 2002). A covariable (LV) × treatment interaction was included in the model statement to determine if treatment differences varied as the LV scores varied across sites. For those LVs that resulted in a significant (P < 0.05) LV score × treatment interaction, means were estimated at the lowest, average, and highest level of the LV scores. X loadings for significant LV interactions were used to infer environment/climate predictors responsible for the covariable × treatment interaction.

A grouping methodology, as previously described by Francis and Kannenberg (1978) , was used to further explore treatment responses among sites. The mean and CV were estimated for each level of the treatment of interest across remaining treatments, sites, and replicates. Means were plotted against CV for each treatment combination, and overall mean of the treatments means and CVs was included in the plot to categorize the data into four categories: Group I: High mean, low variability (optimal); Group II: High mean, high variability; Group III: Low mean, high variability (poor); and Group IV: Low mean, low variability.


Soil and Environmental Conditions

The soil residual nutrient levels in the soil before seeding at each location are presented in tbl2Table 2. A wide range of fields with Cl and K levels were found. No fields were found that were low in K and high in Cl. Growing season precipitation and average temperatures are presented in tbl3Table 3. Precipitation was more than 10% above average at Carry the Kettle in 2007 and Stewart Valley in 2008 and precipitation was at least 10% below average at Indian Head in 2008 and 2009, Stewart Valley in 2007 and 2009, and Regina plain in 2007 and 2009. At all locations, the average temperature for the growing season was below110% of the 30-yr average for that site. On the other hand, the temperature was 90% or less of the 30-yr average for several locations and years including Indian Head in 2009, Carry The Kettle in 2008 and 2009, and the Regina plain in 2009. During this study, temperature tended to be near average or below average at the sites where the experiment was conducted.

View Full Table | Close Full ViewTable 2.

Soil residual levels of Cl, K, N, P and S at the start of the each field experiment.

Year Location Cl
0–15 cm 0–60 cm 0–15 cm supply rate 0–15 cm supply rate 15–30 cm 0–60 cm 0–15 cm 0–15 cm
kg Cl ha−1 kg K ha−1 g K cm−2 kg N ha−1 kg P ha−1 kg S ha−1
2007 Stewart Valley 18.3 76.3 466 10.8 5.0 37.8 41.2 6.6
CTK ellisboro 15.7 57.3 155 5.9 4.0 25.5 29.6 5.2
CTK loam 14.7 49.3 246 8.4 4.8 36.9 19.4 5.6
Regina plain 16.0 84.2 528 4.9 3.6 54.9 19.0 22.3
2008 Indian Head 90.8 242 508 4.0 1.3 79.3 10.9 11.1
CTK ellisboro 36.2 168 207 8.3 3.1 26.9 27.7 5.3
CTK loam 47.8 224 310 7.2 2.6 56.6 21.7 7.7
Regina plain 70.4 376 512 3.1 2.0 31.1 5.4 6.4
2009 Indian Head 72.9 215 717 5.4 3.3 28.7 25.7 52.2
Stewart Valley 66.4 213 465 8.7 2.7 50.2 54.8 66.4
CTK ellisboro 42.6 176. 303 10.6 2.7 28.1 25.6 4.2
CTK loam 42.7 138 303 8.3 2.3 52.9 21.3 17.1
Regina plain 44.7 358 576 3.0 1.6 39.8 7.5 6.2

View Full Table | Close Full ViewTable 3.

Summary of climatic conditions and percentage of the 30-yr long term average from May first to August 31st for experimental sites in Saskatchewan in 2007–2009.

Location/year Precipitation
Percent of 30-yr long-term average Air temperature
Percent of 30-yr long-term average
May June July August May June July August
mm °C
Indian Head
 2008 21 60 90 47 86 9 14 17 18 92
 2009 15 60 59 77 83 8 14 14 15 81
 Long-term average 56 79 67 53 11 16 18 18
Stewart Valley
 2007 67 53 34 29 72 12 16 23 18 106
 2008 98 138 91 97 168 13 15 18 18 98
 2009 12 51 61 61 73 10 15 17 17 91
 Long-term average 56 76 69 52 12 16 19 18
Carry the Kettle
 2007 83 47 51 65 110 11 15 20 16 98
 2008 7 81 96 39 99 9 14 17 13 81
 2009 20 57 42 105 100 8 14 14 15 81
 Long-term average 43 87 49 45 11 16 18 18
Regina plain †
 2007 63 24 22 49 67 11 16 22 16 100
 2008 8 69 75 80 99 10 13 18 18 91
 2009 38 50 45 78 90 9 16 16 16 88
 Long-term average 53 75 64 43 12 16 19 18
Bratts Lake is the closest environmental station to this location.

Crop Development and Seed Yield

When the data was averaged across all locations the application of K or Cl had no effect on the plant density or panicle density of annual canarygrass (tbl4Table 4) and plant density and panicle density were at a level that would not restrict the yield potential of annual canarygrass (tbl5Table 5). Plant height was not affected by the application of K or Cl. The height of the annual canarygrass was between 86 and 87 cm for all treatments. The NDVI was measured before the flag leaf was visible and the application of K or Cl had no effect on NDVI. The NDVI of the annual canarygrass ranged from 0.4537 to 0.4925 for all 10 treatments. The major disease that affects annual canarygrass, septoria leaf mottle (Septoria triseti), was not affected by K or Cl. There were no visible differences observed from the application of K or Cl before heading. In winter wheat, Engel et al. (1994) observed that there were visible differences in plant development between the boot and flowering stages.

View Full Table | Close Full ViewTable 4.

Statistical analysis of all sites combined.

Source Plant density Head density Panicle size Seed density Kernel weight Grain yield Height NDVI† Test weight
plants m−2 panicles plant−1 panicles m−2 seeds panicle−1 seeds m−2 g 1000 kernels−1 kg ha−1 cm 0–1 g 0.5 L−1
P value
Treatment 0.880 0.466 0.888 <0.001 <0.001 0.079 <0.001 0.703 0.261 0.081
K × Cl 0.890 0.546 0.907 0.832 0.661 0.594 0.674 0.562 0.849 0.951
Cl rate 0.288 0.104 0.593 0.009 0.001 0.439 0.001 0.405 0.152 0.104
Cl linear 0.192 0.043 0.382 0.015 0.002 0.584 0.004 0.716 0.570 0.052
Cl quadratic 0.189 0.266 0.293 0.076 0.022 0.137 0.015 0.311 0.066 0.119
K rate 0.583 0.092 0.412 0.807 0.763 0.809 0.739 0.578 0.213 0.747
K linear 0.185 0.054 0.099 0.538 0.937 0.916 0.860 0.424 0.479 0.524
K quadratic 0.675 0.107 0.772 0.686 0.681 0.416 0.806 0.250 0.082 0.691
Variance estimate
Site (S) 4,789.7 0.7778 20,489.7 167.0 70,713,785 0.1587 432529.9 578.7 0.0178 260.1
Treatment × S 130.9 0.0256 -23.1 5.942 2,586,908 0.0221 20266.5 -0.5059 -1.9E-05 657.9
Variance estimate P value
S 0.017 0.018 0.016 0.016 0.015 0.022 0.015 0.015 0.050 0.074
Treatment × S 0.149 0.201 0.932 <0.001 <0.001 0.011 <0.001 0.655 0.910 <0.001
Percentage of total variance
Treatment × S§ 3 3 0 3** 4** 12* 4** 0 0 72**
*Significance at 0.05 probability level.
**Significance at the 0.01 probability level.
NDVI, normalized difference vegetative index.
P values were bolded when they were less than 0.05.
§This is the percentage of the total variance that is associated with effect of site. Treatment × S variance estimate/(Treatment × S variance estimate + S variance estimate) × 100.

View Full Table | Close Full ViewTable 5.

The effect of Cl and K on the development and seed yield of annual canarygrass.

Fertilizer treatment Plant density Panicle density Panicle size Seed density Kernel weight Seed yield Test weight
kg K and Cl ha−1 plants m−2 panicles plant−1 panicles m−2 seeds panicle−1 seeds m−2 g 1000 kernels−1 kg ha−1 g 0.5 L−1
0K-0Cl 260 2.7 581 21.7 12,641 7.81 991 315
10K-9.1Cl 274 2.3 597 25.3 15,428 7.95 1227 338
20K-18.2Cl 272 2.4 594 25.6 15,020 7.91 1189 339
30K-27.3Cl 267 2.5 600 26.0 15,561 7.99 1231 342
10K-0Cl 268 2.4 589 20.8 12,159 7.74 952 315
20K-0Cl 274 2.4 599 21.4 12,865 7.79 1015 326
30K-0Cl 274 2.4 608 20.6 12,381 7.81 961 319
0K-9.1Cl 268 2.4 590 25.8 15,384 7.94 1226 341
0K-18.2Cl 276 2.4 600 25.2 15,216 7.93 1202 341
0K-27.3Cl 269 2.4 597 26.1 15,758 7.87 1235 340
LSD 0.05 ns† ns ns 2.6 1,468 ns 126 ns
ns, not significant.

Panicle size (seeds panicle−1) and seed density (seeds m−2) were affected by the application of Cl but not by K (Table 4). Panicle size (seeds panicle−1) showed a linear response to increases in rate of Cl and seed density (seeds m−2) showed a curvi-linear response with increasing rates of Cl. The form of Cl used did not affect the response of annual canarygrass to Cl (Table 5). Kernel weight was not affected by the application of Cl or K. Therefore, the main yield component that was affected by the addition of Cl was panicle size (seeds panicle−1). The Cl appears to have helped the annual canarygrass fertilize more ovules or retain seed during the reproductive development of annual canarygrass. A similar response was observed in a specific hard red spring wheat cultivar, Marshall, with an increase in seed density and not kernel weight (Evans and Riedell, 2006). This cultivar appears to take up less Cl than other wheat cultivars at a given Cl concentration in sand culture (Evans and Riedell, 2006). Other research found that kernel weight and not seed density was the yield component that was increased in cereals (Gaspar et al., 1994; Engel et al., 1994).

There was a curvi-linear increase in seed yield to the application of Cl but not K (Table 4). Averaged across all sites, seed yield increased by approximately 24% when Cl was added in the form of KCl or CaCl2 (Table 5). The lowest Cl rate used in the study, 9.1 Cl kg ha−1 was sufficient to increase yield and no responses to higher rates of Cl were observed. A response curve cannot be accurately calculated since the response occurred between two rates 0 and 9.1 kg Cl ha−1. In previous research reporting seed yield responses to Cl, yields tended to increase over a wider range of Cl rates (Fixen et al., 1986; Engel et al., 1994).

The form of Cl did not matter, with both KCl and CaCl2 being equally effective in increasing seed yield (Table 5). This lack of sensitivity to the form of Cl has also been observed in wheat (Fixen et al., 1986; Mohr et al., 1995b) and barley (Mohr et al., 1995b). When the means of seed yield, panicle size (seeds panicle−1) and seed density (seeds m-2) were plotted against the coefficient of variation it became clear that all the rates of Cl evaluated increased yield and reduced yield variability (fig1Fig. 1). Test weight followed a similar trend as seed yield; however the linear contrast had a P value of 0.052 (Tables 4 and 5).

Fig. 1.

Biplot [estimated means vs. coefficient of variation (CV)] of the 10 treatments for data collected from six sites over 3 yr in Saskatchewan from 2007 to 2009. Group I: High mean, low variability (optimal); Group II: High mean, high variability; Group III: Low mean, high variability (poor); Group IV: Low mean, low variability.


The interaction between the treatments and site was significant for seed density, kernel weight, seed yield, and test weight (Table 4). Seed density (seeds m−2) deviated from the overall means at CTK loam 2008, CTK ellisboro 2009, Indian Head 2008, and Regina plain 2008. For the two CTK site-years, the response to Cl was larger than the over all response while at the other two locations the response to Cl was smaller than when averaged over all the sites (tbl6Table 6). This site response to Cl for seed density (seeds m−2) resulted in three sites, CTK loam 2008, CTK ellisboro 2009, and Regina plain 2008, having seed yield treatment means which deviated from the overall treatment means. The seed yield response to Cl was larger at CTK loam 2008 and CTK ellisboro 2009 and smaller at Regina plain 2008. CTK ellisboro 2009, the site-year with the largest response to Cl was the only location where all treatment means for this site deviated from the overall treatment means. At this site, the addition of Cl increased test weight (data not shown). It is interesting to note that this site did not have the lowest level of Cl (Table 2). When the site × treatment interaction for kernel weight was examined, no clear trend emerged with just one site mean in 1 yr deviating from the overall means (data not shown).

View Full Table | Close Full ViewTable 6.

Deviation of the site means from the overall mean of seed density, seed yield, and test weight.

Location/year Treatments
0K-0Cl 10K-9.1Cl 20K-18.2Cl 30K-27.3Cl 10K-0Cl 20K-0Cl 30K-0Cl 0K-9.1Cl 0K-18.2Cl 0K-27.3Cl
Seed density (no. m−2)
12,641 15,428 15,020 15,561 12,159 12865 12381 15,384 15,216 15,758
(Deviations; no. m−2)
CTK loam 2007 –1,397 364 1,348 776 –900 –1659 –1520 594 1,241 852
CTK loam 2008 –2,222* 3,441** 829 –1,537 –1,202 –308 –2686** 1,911 262 1,847
CTK loam 2009 646 –938 –321 –432 583 948 –193 –801 329 446
CTK ellisboro 2007 603 –281 –511 –694 990 533 754 –678 –357 –848
CTK ellisboro 2008 –1371 517 591 484 –961 –1157 –950 806 1258 404
CTK ellisboro 2009 –3336** 1934 2,664** 1,924 –3593** –3599** –2516* 2198* 1,883 2,255*
Indian Head 2008 2,024* –1521 –167 189 429 1947 1371 –949 –2,105* –860
Indian Head 2009 1021 –2006* –340 –854 947 582 1834 –1,126 269 141
Regina plain 2007 763 48 –979 570 913 209 464 –955 –549 –615
Regina plain 2008 2,144* 126 –362 –171 874 292 527 159 –967 –2,470*
Regina plain 2009 756 –508 –1696 –163 1,066 1110 839 –820 –800 354
Stewart Valley 2007 807 –623 –675 –661 926 1353 741 –339 –853 –786
Stewart Valley 2009 –438 –553 –381 569 –74 –250 1334 –1 388 –719
Seed yield (kg ha−1)
991 1227 1189 1231 952 1015 961 1226 1202 1235
(Deviations; kg ha−1)
CTK loam 2007 –132 28 124 83 –85 –158 –136 47 112 88
CTK loam 2008 –157 286** –39 –141 –101 –44 –303** 289** –20 264**
CTK loam 2009 67 –86 0 –46 47 79 –4 –59 –26 60
CTK ellisboro 2007 55 –32 –48 –61 88 44 75 –69 –33 –68
CTK ellisboro 2008 –103 21 24 54 –72 –102 –61 54 114 32
CTK ellisboro 2009 –307** 170* 254** 202* –331** –334** –216* 171* 172* 203*
Indian Head 2008 147 –70 4 –36 72 143 124 –97 –130 –129
Indian Head 2009 87 –145 –11 –84 75 137 125 –65 32 –99
Regina plain 2007 78 11 –82 75 53 20 54 –109 –64 –50
Regina plain 2008 180* –13 –34 –51 84 39 69 –31 –49 –182*
Regina plain 2009 57 –75 –105 –4 110 104 105 –86 –78 –17
Stewart Valley 2007 73 –49 –60 –54 81 107 51 –38 –69 –54
Stewart Valley 2009 –44 –46 –27 62 –21 –35 118 –7 38 –48
*Significance at 0.05 probability level.
**Significance at the 0.01 probability level.

Soil Chloride Levels and Seed Yield

To determine the suitability of using soil test measurements for recommending the application of Cl, a partial least squares analysis was performed (principal component analysis with a regression component added). Variable importance for the projection was generated for Cl in the 0- to 15-cm depth, of 2.01 and for Cl in the 0- to 60-cm depth of 1.68. This indicated that the 0- to 15-cm depth was more consistent in predicting the responsiveness of annual canarygrass to the application of Cl. A sensitivity analysis was performed indicating that once the level of residual Cl in the soil reached 70 kg ha−1 in the 0- to 15-cm depth, the effect of applying Cl became insignificant (P = 0.074; fig2Fig. 2). Therefore, soils with levels below 70 kg Cl ha−1 in the 0- to15-cm depth should receive Cl when annual canarygrass is grown. More data would improve the accuracy of this recommendation. This may indicate that annual canarygrass is more sensitive to a Cl deficiency than other cereal crops. Fixen (1993) found that soil Cl levels needed to be below 50 kg ha−1 in a 0- to 60-cm depth to elicit a seed yield increase in wheat with the application of Cl fertilizer, while this study indicates that Cl fertilizer should be applied when Cl levels drop below 70 kg ha−1 in the 0- to 15-cm depth. In addition, at the site with the largest seed yield response to Cl in annual canarygrass, CTK ellisboro, no response to Cl was found in wheat when preliminary studies on wheat and annual canarygrass were conducted in the same year (Lafond, 2001; May, 2002). Annual canarygrass may be a suitable species to use to investigate the effects of Cl on reproductive development in plants.

Fig. 2.

Sensitivity analysis of seed yield in annual canarygrass to soil Cl levels in a 0- to 15-cm depth. Yield effect is the difference between the average seed yield of treatments 8, 9, and 10 vs. entry 1. Error bars that cross horizontal axis indicate those effects/differences not different from zero.



The response of annual canarygrass seed yield to Cl occurred under low or high yielding conditions. The yield component that was responsible for the yield increase was panicle size (seeds panicle−1). No response was observed to K even in fields where the residual levels were low to moderate in K according to soil test extractable K and soil K supply rate. Annual canarygrass growers are advised to measure Cl when doing soil tests. It is recommended that 9.1 Cl kg ha−1 in the form of 20 kg ha−1 of KCl be applied when the Cl level in the soil (0–15 cm) is below 70 kg Cl ha−1. Annual canarygrass appears to be more sensitive to Cl deficiency than wheat.


This research was funded by the Canaryseed Development Commission of Saskatchewan, Indian Head Agricultural Research Foundation, Agriculture and Agri-Food Canada, Saskatchewan Agriculture and Food, International Plant Nutrition Institute, and Agrium. I would like to thank Craig Stevenson for his assistance with the statistical analysis. The technical support of Orla Willoughby, and Randy Shiplack at the Indian Head Research Farm, Darwin Leach at the Melfort Research Farm, and Don Sluth from the Wheatland Conservation Association was appreciated.




Be the first to comment.

Please log in to post a comment.
*Society members, certified professionals, and authors are permitted to comment.