Optimizing Lentil and Pea Agronomy for Organic Production. Final Report April 10, 2009

Transcription

1 Optimizing Lentil and Pea Agronomy for Organic Production Final Report April 10, 2009 Prepared for SPG: AGR0504 Principal Investigators: Dr. S. Shirtliffe Department of Plant Sciences 51 Campus Drive University of Saskatchewan Saskatoon, Saskatchewan S7N 5A8 Phone: (306) Dr. F. Walley Department of Soil Science Phone: (306) Co-investigator: Dr. Diane Knight Department of Soil Science Phone (306) Prepared by: Steve Shirtliffe, Julia Baird, Boldsaikhan Usukh and Fran Wally

2 1 Abstract There are no seeding rates established for organic production of field pea and lentil in Saskatchewan and organic producers must rely upon rates recommended for conventional production of these crops. These seeding rates may not be suitable for organic production as the two systems differ in the use of inputs and in pest management. The objectives of this study were to determine an optimal seeding rate for organic production of field pea and lentil in Saskatchewan considering a number of factors, including yield, weed suppression, soil nitrogen (N) and phosphorus (P) concentrations, soil water storage, colonization of crop roots by arbuscular mycorrhizal fungi (AMF), plant P uptake, soil nutrient dynamics, rotation effect on subsequent crops and profitability. A field experiment was conducted to determine the optimal seeding rates of field pea and lentil. Seed yield increased with increasing seeding rate for both crops while weed biomass decreased. Post-harvest soil phosphate-p levels did not change consistently between treatments, indicating that there was no trend in soil P concentration with seeding rate. Post-harvest soil inorganic N, however, was higher for the summer fallow and green manure treatments than for the seeding rate treatments in both crops. Soil water storage following harvest was not affected by treatment. Colonization of crop roots by AMF increased for lentil with increasing seeding rate, but the same trend was not observed in field pea. However a growth chamber experiment to study the rate of colonization of field pea did not show any differences in AMF colonization between seeding rates. Arbuscular mycorrhizal fungi colonization and seeding rate had no effect on plant P concentration for either field pea or lentil. Seeding rate did not affect the percentage of nitrogen from fixation for both crops, however the amount on nitrogen fixed increased with increased seeding rate because of the increase in crop biomass. Although there was a positive nitrogen balance from growing the grain legumes at high seeding rates, there was no positive effect on subsequent wheat yield. In contrast, growing a green manure crop or summer fallowing resulted in yield increase in wheat. This effect was not due to water storage as there was no difference in soil water levels between the treatments.

3 Both crops became increasingly profitable as seeding rate increased. Field pea reached a maximum return at 200 seeds m -2 and lentil return increased to the highest seeding rate of 375 seeds m -2. However the maximum return was dependent on the seed cost and was quite flat near the maximum rate of return. Thus a seeding rate of 150 seeds 150 seeds m -2 for field pea and 250 seeds m -2 for lentils is recommended. Growers who anticipate lower than normal yields, have high seed costs or who are adverse to financial risk can seed at a lower rate but with lower net returns expected under average conditions. 2 Executive Summary Please see the above abstract. 2.1 Background The overall objective of this research project was to determine the optimum seeding rate for lentil and field pea in Saskatchewan. To do this, two multi-year experiments were conducted. In the first year the effect of seeding rate was established in either pea or lentil. The following year the plots were re-cropped to wheat to access the rotation effect of these practices. Green manure crops and summer fallow were included as controls. We used a multi-disciplinary approach that examined aspects of agronomy, weed control, root symbionants, soil nutrient cycling and economics. This experimental report is arranged in five sections; the first two sections cover the effect of the seeding rate on the pea and lentil in the year of production, the third section examines the effect of seeding rate on mychorrhial infection using both a field and growth chamber experiment the forth section examines the effect of seeding rate in pulse on nitrogen fixation and the yield of the subsequent wheat crop. Finally the last section validates the seeding rate recommendations developed in the first section.

4 3.1 Experiment 1 Optimal seeding rate for organic production of field pea in the northern Great Plains Introduction and Objective: The province of Saskatchewan supports the largest number of certified organic farms in Canada, and produces the vast majority of organic field pea (Pisum sativum) (>10,000 ha) grown in the country (Macey 2006). Field pea is known to be poorly competitive with weeds (Wall et al. 1991; Wall and Townley-Smith 1996; Harker 2001). Yield losses due to weed interference can be as high as 80% (Boerboom and Young 1995; Grevsen 2003). In addition, Leeson et al. (2000) found that organic farms had a higher number of weeds after post-emergent weed control than conventional farms. Increasing the crop density has been shown to reduce weed densities (Townley- Smith and Wright 1994), and is commonly used in organic agricultural systems as a cultural weed management strategy together with other practices such as post-emergence tillage and crop rotation (SOD 2000; Stockdale et al. 2001; Nazarko et al. 2003). While increasing seeding rate may increase the competitive ability of the crop, the profitability of the crop may or may not increase, as the cost of the seed must be taken into consideration. Pulse seed in particular can be prohibitively expensive when increasing seeding rates (Loss et al. 1998). For example, the price paid for organic field pea ranged from $0.88 to $1.98 per kg in 2005 (University of Saskatchewan 2006). Various methods can be used to determine economic optimum seeding rate. The most common method was developed by French et al. (1994), where non-linear regression was fitted to describe the yield-density response, and then the slope where the optimal return was reached was determined and the ideal density identified. Shirtliffe and

5 Johnston (2002) fit non-linear regression equations to dry bean gross return (minus seed cost) and density. Economic optimum plant densities were determined from the density where return was maximized. The current seeding rate recommendation for field pea has been determined for conventional management systems. Saskatchewan Pulse Growers (2000) suggest a target stand of 88 plants m -2 in western Canada. However, the seeding rate to optimize production of this crop in organic systems has not been determined for this region. An organic production study in France found that field pea populations greater than 80 plants m -2 decreased weed densities (Corre-Hellou and Crozat 2005). Because organic and conventional production systems are very different in terms of pest and nutrient management, seeding rates established for conventional production may not be suitable for organic systems. Increasing the seeding rate of field pea may increase its competitive ability and decrease yield loss from weeds. Increasing seeding rates has been shown in many instances to increase yield and competitiveness of the crop (Wall et al. 1991; Mohler 2001); however, the effects on soil fertility and seed quality are varied or unknown. The objective of this study was to determine the optimal seeding rate for organic production of field pea based on profitability as related to a number of other factors including yield, weed management and soil fertility.

6 3.1.2 MATERIALS AND METHODS Experimental Design and Plot Management Two locations were chosen on pre-existing organic farms in The locations were south of Vonda, SK and west of Delisle, SK. The sites were certified organic for approximately 20 and 8 yr, respectively. Both sites were seeded to barley (Hordeum vulgare) the previous year. In 2006, two sites were chosen near Vonda, SK and Vanscoy, SK. The sites were certified organic for approximately 20 yr and 3 yr, respectively. Both sites were seeded to wheat (Triticum aestivum) the previous year. Site locations and descriptions are given in Table Precipitation and temperature data were recorded during the growing season by weather stations set up in close proximity to the research plots. Field pea (Pisum sativum cv. Mozart) was grown in a randomized complete block design with five target plant densities of 10, 25, 62, 156 and 250 plants m -2 and two fallow treatments (green manure and cultivated fallow) with four replicate blocks. The fallow treatments were included to compare the soil fertility relative to the seeding rate treatments. Trapper field pea was used as the green manure crop. The experiment was conducted over two years with two sites each year with a single plot size of 2 m x 6 m. Field peas were seeded at a 23 cm row spacing 2.5 cm deep using a cone seeder with an offset disc drill. The number of seeds planted was increased based on the percent germination derived from a germination test to achieve target plant densities. When seeding, Nodulator granular Rhizobium inoculant (Becker Underwood, Saskatoon, SK) was placed with the seed at the recommended rate. The natural weed population was used in this study, as the population was relatively homogeneous.

7 Trapper field pea was seeded at a rate of 62 viable seeds m -2 (Lawley 2004). An additional green manure treatment was seeded at a rate of 156 viable seeds m -2 in 2006 in an effort to reach the target stand density of 62 plants m -2. Pre-plant tillage was performed at all sites to manage early-emerging weeds. Incrop harrowing was performed with two passes using a tine harrow parallel to the crop when it was judged to be resilient enough to withstand the treatment (approximately one month after seeding). Tillage (two passes using a tandem disc) was performed in the summerfallow treatment to manage weeds as required. Actual crop and weed species densities were determined by counting the number of each species in two randomly selected 0.25 m 2 quadrats in each plot approximately one week after the in-crop harrowing was performed. The optimal crop stage for the green manure ploughdown was the early bud stage (Lawley 2004); however, the green manure ploughdown often occurred later than the optimal stage due to inclement weather. The ploughdown was performed in two passes using a tandem disc. A second tillage pass was performed at a later date for both the summerfallow and green manure treatments to manage weeds. Aboveground biomass samples of both crop and weeds were taken at two random locations within each plot at physiological maturity (indicated by yellow pods) and combined for each plot. Each sample was 0.25 m 2 and was cut one cm above ground level. Plant material was separated into crop and weed species. All plant material was dried at 60 C for 72 h and then weighed to determine biomass. An aboveground sample of four-1 m long rows (equivalent to 0.81 m 2 ) was also taken at harvest from each plot to determine seed yield and harvest index.

8 Soil Analysis Soil samples were taken at depths of 0 to 15 cm and 15 to 30 cm at five random locations within each trial area just prior to seeding. Samples from each depth were bulked and sent to ALS Laboratory Group (Saskatoon, SK) for ph, electrical conductivity (EC), and macronutrient analysis (Table 3.1.1). Two soil samples were taken after harvest in each plot in randomly selected locations at depths of 0 to 15 cm and 15 to 30 cm and samples at each depth were bulked within each plot. Each bulked sample was subsampled, air-dried and then ground to pass through a 2 mm sieve. Subsamples were taken for determination of available soil nutrients. Phosphate-P at the 0- to 15-cm depth was measured using the modified Kelowna extraction method (Qian et al. 1994). Inorganic N (NH NO - 3 ) was measured by KCL extraction for both the 0- to 15-cm and 15- to 30-cm depths (Maynard and Kalra 1993) by ALS Laboratory Group (Saskatoon, SK) Plant Analysis Aboveground biomass sampled at harvest was air-dried in a covered outdoor facility in cloth bags for one week, then moved indoors and dried further in a heated facility at approximately 30 C. Samples were stored indoors until threshing. Samples were weighed prior to threshing, and then threshed by machine. Seed weight was determined from the cleaned, threshed seed and vegetative weight was determined by subtracting seed weight from the total biomass sample weight. Seed quality was assessed using the Canadian Grain Commission s seed grading guides for field peas (CGC 2005). A further assessment of seed quality using 100 seed weight was also performed. Two samples of 100 seeds were weighed and averaged for each plot.

9 Incidence of disease was a concern for denser stands, so grain samples from the highest seeding rate were tested for seed-borne disease. Fifty seeds from each replicate were surface sterilized with a 10% bleach solution for two minutes, then plated using sterile technique onto potato-dextrose agar (10 seeds per Petri dish) and set onto lighted benches for one week (Morrall and Beauchamp 1988). Assessments of occurrence of disease were made by a plant pathology technician based on the colour and morphology of the colonies observed. After seed quality and disease incidence were assessed, remaining seed and straw samples were ground and stored for nutrient analysis. Both seed and straw samples were assessed for P content by performing an acid digestion (Thomas et al. 1967). Phosphorus levels in the digested samples were determined colorimetrically using a Technicon AutoAnalyzer II (Technicon Industrial Systems, Tarrytown, NJ) Statistical Analysis Yield data were analyzed using the PROC MIXED procedure of SAS (SAS Institute Inc. 2004) with block and site-year as random effects and seeding rate as the fixed effect. Treatment effects were considered significant at P Data were tested and transformed to meet the assumption of homogeneity of variance. The transformations used are listed in Table Back-transformed data are presented. Sites and years were combined as there were no significant interactions between site-year and treatment (data not shown). Significant results are presented in-text with the F value, numerator and denominator degrees of freedom and P value enclosed by brackets. The PROC GLM procedure of SAS (SAS Institute Inc. 2004) was to perform linear and quadratic contrasts to determine significant trends in data other than yield data,

10 and PROC GLM was also used to compare means between summerfallow, green manure and the highest seeding rate treatments. A non-linear regression model was fitted to the crop biomass and grain yield data using the PROC NLIN procedure of SAS (SAS Institute Inc. 2004). The two-parameter Michaelis-Menten model was used with a modification of parameters: [2] where Y max (kg ha -1 ) is the maximum pea yield reached as the density approaches infinity, D 50 (viable seeds m -2 ) is the crop density at which half of Y max occurs, and D (viable seeds m -2 ) is seeding rate. The D 50 parameter describes the slope of the curve. This model was chosen to describe the data because it provided a good fit to the data and biologically meaningful parameters. The relationship between weed biomass and crop density was described using a modified version of the Michaelis-Menten equation to fit decreasing rather than increasing biomass: [3] The fit of the equation to the means was determined using the adjusted R 2 value in SigmaPlot 10.0 (Systat Software, Inc.) Economic Analysis To determine the profitability of each seeding rate, an equation was used to calculate the potential return (Norsworthy and Oliver 2001): [4] where R is return ($ ha -1 ), Y is seed yield (kg ha -1 ), P is price received, SW is seed weight planted (kg ha -1 ) and C is seed cost ($ kg -1 ).

11 A modified Michaelis-Menten equation was fitted to the economic return data for the four site-years combined. This equation is algebraically equivalent to the yielddensity equation by Firbank and Watkinson (1990). This re-parameterized equation accounted for the decrease in return at the highest seeding rate: [5] where R max is the maximum return ($ ha -1 ), D is the seeding rate (viable seeds m -2 ), D 50 is the seeding rate at which half of the maximum return value would be reached (viable seeds m -2 ) and c is a constant. The optimal seeding rate for each crop was determined empirically the maximum return was reached, as the parameters used in this equation were not biologically meaningful. The cost of seed was provided by an organic seed supplier near Saskatoon, SK (Marysburg Organic Producers, Inc., personal communication). The seed cost used was $0.27 kg -1. The seed weight planted was determined from the 1000 seed weight counted and weighed prior to seeding multiplied by the number of seeds planted for each seeding rate. The number of seeds planted depended upon the percent germination, as previously described. The gross profit was determined from the average selling price for each crop in Saskatchewan from 2005 as reported by the University of Saskatchewan (2006). The average selling price used was $0.23 kg -1.

12 RESULTS AND DISCUSSION Weather Climatic conditions varied between years and from the 30 yr average for the Saskatoon area (Table 3.1.4). Crop Emergence and Growth Low establishment rates resulted in lower plant densities than intended (Fig ). Stand establishment increased linearly as seeding rate increased (slope = 0.6, R 2 =0.99). Establishment rates ranged from 52% to 66% with no trend as seeding rate increased. Field pea biomass at physiological maturity increased asymptotically with increasing seeding rate (Fig ). Biomass increased from 602 to 4695 kg ha -1 between 10 and 250 viable seeds m -2. The increase in biomass was significantly related to seeding rate (F5, 33.6 = , P < 0.001). One half of the predicted maximum biomass yield was reached at a seeding density of 110 viable seeds m -2 (Table 3.1.5). As crop density increased, weed biomass decreased (Fig ). Seeding rate effects were significant for weed biomass reduction (F5, 33.8 = , P < 0.001). Weed biomass was reduced by half at 92 viable seeds m -2, the same seeding rate at which grain yield reached half of the predicted maximum (Table 3.1.5). Total weed numbers did not decrease significantly with increasing seeding rate (data not shown). Weed biomass was reduced by 68% between the lowest and highest seeding rates (Fig ). These results are consistent with the findings of other studies. Grevsen (2003) found that weed biomass decreased by 30 to 50% when seeding rate of green pea was increased from 90 to 150 plants m -2. Similarly, Boerboom and Young (1995)

13 reported a reduction in weed biomass when crop density was increased by more than 50% above the recommended density of 88 plants m -2 for pea, although the weed density did not decrease. Environmental conditions may have played a role in the competitiveness of the weeds and crop. The temperature and precipitation were higher than normal for most of the growing season in both years (Table 3.1.4) and may have increased the competitiveness of weed species such as wild mustard which was prevalent at Vonda in both years. Grain Yield Field pea seed yield showed a similar trend to crop biomass, increasing asymptotically to the highest seeding rate (Fig ). Yield increased significantly with seeding rate, from 209 to 1725 kg ha -1 between 10 and 250 viable seeds m -2 (F5, 33.8 = , P < 0.001). The trend for increasing crop yield and biomass with increasing seeding rate has been demonstrated in a number of studies. McDonald et al. (2007) also found that pea yield increased to the highest treatment density of 200 plants m -2. Similarly, Lawson (1982) found that pea yields increased with increasing plant density up to 140 plants m -2 in one experiment, and up to 195 plants m -2 in another. Pea biomass production and yield increased with increasing seeding rates in stress conditions (Boerboom and Young 1995). However, Johnston et al. (2002) discovered no increase in seed yield associated with increased seeding rate between 50 and 150 plants m -2 for semileafless pea in one year and only a small increase during another year. The rates are comparable to the rates used in our study, but our results refute those of Johnston et al. (2002), as yield increases occurred between all seeding rates, from 10 to 250 viable seeds m -2.

14 The Michaelis-Menten equation described the biomass and crop yield means well (R 2 =0.92 to 0.99). Typically, the Michaelis-Menten equation is used to describe enzyme kinetics, whereas in this study it was used in a novel manner to describe the effects of increasing seeding rate on crop biomass and yield in a weedy setting, and modified to describe weed biomass reduction with increasing seeding rate. Previously, the Michaelis- Menten equation was adapted by Jolliffe et al. (1984) to describe crop yield for a single species grown without interspecific competition. Other workers have used a variety of equations to describe crop yield trends with increasing seeding rate, including asymptotic (Gooding et al. 2002), linear (McDonald et al. 2007), and quadratic (Lawson 1982) curves. An absolute maximum was not reached for grain yield at seeding rates up to 250 viable seeds m -2. The actual plant density reached at this seeding rate was 146 plants m -2. Other studies have found that conventional seeding rate studies have reached a maximum yield at high seeding rates. For example, Johnston et al. (2002) found that pea yields reached a maximum at approximately 125 plants m -2 with a yield of 2800 kg ha -1 in weed-free conditions. Similarly, Lawson (1982) reported that vined pea yield reached a maximum at 9000 kg ha -1 at a plant density of 160 plants m -2 in one experiment. In a separate experiment, yield was hyperbolic and reached a maximum at 6000 kg ha -1 at a plant density of 120 plants m -2. The yield-density relationship in our study may have reached a maximum if emergence rates had been higher. Another possible hindrance to achieving a maximum yield was the presence of weeds in our study. Most conventional seeding rate studies are performed in weed-free conditions. One exception was a pea seeding rate study by Wall and Townley-Smith

15 (1996) where wild mustard (Sinapis arvensis) was seeded with the crop. Weed emergence 10 d prior to crop emergence corresponded to an increase in the slope of the curve that described the yield-density response as compared to weed emergence 2 d prior to crop emergence. Timing of weed emergence may have also had an effect on pea yields in our study. At most sites, pre-plant tillage occurred several days prior to seeding. In many cases the crop and weeds were similar in size when the in-crop harrow occurred; rendering the harrow less effective than if the weeds had emerged after the crop. Yield Components and Seed Quality Harvest index did not change between seeding rates except a decrease at the highest seeding rate from 0.47 to 0.44 (Table 3.1.6), indicating that more vegetative biomass was produced at the highest crop density than any other, although the trend was not significant. Hundred seed weights did not change significantly between crop densities, indicating that seed quality was maintained (Table 3.1.6). Visual inspection of seed samples from all seeding rates indicated that there was no trend in seed grade between seeding rates, and all samples in both years met the Canadian Grain Commission (CGC) standards of Grade 1. Similar results were obtained by Lawson (1982), where there was no reduction in grain weight in vining pea as seeding rate increased from 11 to 194 plants m -2. Johnston et al. (2002) found no reduction in semi-leafless pea weight at one study site between 50 and 150 plants m -2, but did report reduced seed weight at high crop densities at another site. An assessment of seed-borne disease for the highest seeding rate determined that incidence of disease was rare at all sites (Stephanie Boechler personal communication).

16 The most common disease-causing agent found was Alternaria where 2.4% of seeds tested were infected. Soil Fertility Seed and straw P concentrations did not change significantly between seeding rates. Although not statistically significant, straw P concentration showed a tendency to increase as seeding rate increased (Table 3.1.7). A significant linear trend occurred for both seed and straw P uptake as a result of increasing biomass (Table 3.1.7). Phosphorus uptake by field pea was much lower in our organic trial as compared to a conventional study where no N fertilizer was added (Deibert and Utter 2004). Seed P uptake in the conventional experiment averaged 8.6 kg ha -1 at harvest and Deibert and Utter (2004) stated that variation was due to biomass production rather than seed P concentration. Our results were lower with seed P uptake of 3.8 kg ha -1 (Table 3.1.8). Although little information was found regarding field pea straw P concentration or uptake, it is likely that straw P uptake would also be lower for our study when compared to conventional P uptake levels. Available P remaining in the soil after harvest varied between sites, but differences within sites were small (Table 3.1.8). These results are not unique; Gosling and Shepherd (2005) found that there was no difference in extractable P concentration between low and high fertility phases of organic farms. When available soil P concentrations were compared with prior spring test results, all sites showed a decrease in available soil P at harvest (Table 3.1.1; Table 3.1.8). Measurable changes occurred in soil inorganic N (NH NO - 3 ) between treatments (Table 3.1.8). Trends were similar between sites, with the highest inorganic N in the summerfallow and green manure

17 treatments. This trend was evident in both the 0- to 15-cm and 15- to 30-cm depths (Table 3.1.8). The differences between summerfallow and green manure treatments as compared to the highest seeding rate were significant (P<0.05) in 2006 but not in 2005, with the exception of a significant difference between summerfallow and the highest seeding rate for the top 15 cm in Delisle (data not shown). The lowest concentrations of soil inorganic N were generally found at the lowest three seeding rates. The data may suggest a trend towards increasing inorganic N levels at higher seeding rates. There were no significant differences in soil inorganic N between green manure and summerfallow treatments at harvest (data not shown), however, much of the N added by the green manure ploughdown may have still been in the organic form whereas much of the N in the summerfallow treatment was likely decomposed and in the inorganic form measured. Inorganic N levels the following spring may have shown a greater inorganic N benefit of growing pea as a green manure rather than employing summerfallow as a soil fertilitybuilding technique. Economics Potential return was significantly related to seeding rate (F5, 7.52 = , P < 0.001), however economic return increased to the second-highest seeding rate then decreased by approximately $2 ha -1 between seeding rates of 156 and 250 viable seeds m -2, although the two amounts were very similar (Fig ). There was no significant site-year x treatment interaction for any of the crop variables measured (data not shown). Therefore, we were able to determine one seeding rate for all sites, rather than a range, as suggested in several studies for various crops (French et al. 1994; Siddique et al. 1998; Jettner et al. 1999).

18 Potential return peaked at $271 at a seeding rate of 200 viable seeds m -2 ; this seeding rate may be considered to be the optimal seeding rate for organic production of field peas in the northern Great Plains. However, emergence rates were much lower than the seeding rates, and the actual plant density achieved at the recommended seeding rate was 120 plants m -2 (Fig ). At this plant density, the yield was 1600 kg ha -1 and seed cost was approximately $118 ha -1. The seed cost increased by $56 ha -1 between the two highest seeding rates and weed biomass was reduced from 1000 to 850 kg ha -1. Grevsen (2003) suggested that seeding rates should be higher than the recommended rate for conventional production. Grevsen (2003) found that the recommended density of 120 plants m -2 for organic pea production in Denmark was a minimum seeding rate, and that seeding rates should be as high as is economically feasible. Our study suggests a seeding rate of 200 plants m -2 to achieve a crop density of 120 plants m -2 which is similar to the results obtained by Grevsen (2003). This optimal organic seeding rate is 112 viable seeds m -2 higher than the recommended target density for conventional production of 88 plants m -2. The increase in seeding rate provided an additional 650 kg ha -1 in weed biomass reduction and approximately $75 ha -1 more in profit. While the seeding rate was much higher than the recommended rate for conventional production, the actual plant density achieved was only 32 plants m -2 higher, and this target plant density may be used by organic farmers with seeding rates adjusted to accommodate reduced germination and emergence. Seed cost may be prohibitive for very high seeding rates of legumes (Siddique et al. 1998). Indeed, seed costs for organically grown pea are generally higher than selling prices, as seed cost includes the cost of seed cleaning. In 2005, seed costs reached $0.27

19 kg -1 (Marysburg Organic Producers, Inc.) and organic selling prices ranged from $0.22 to $0.26 kg -1 in 2005 (University of Saskatchewan 2006) Thus, although the economic return may be greater at high seeding rates, organic growers may find that spring seed costs constrain their decision to increase seeding rate CONCLUSION Increasing seeding rates of field pea will increase crop biomass production and yield,. Weeds in organic production systems can be suppressed by increasing seeding rates. Seed quality and plant P uptake were not significantly affected by seeding rate and are not a constraint to increasing seeding rates. At harvest, inorganic soil N was lower in the grain legume treatments than the summerfallow and green manure treatments. When inorganic N was examined within the grain treatments, however, the highest concentrations occurred in the higher seeding rates. Inorganic N levels may also increase for the following crop as residues decompose. This is encouraging for organic producers who wish to increase seeding rates to manage weeds, but are concerned about soil N losses. Organic farmers will benefit from increasing seeding rates beyond the recommended rate for conventional production to a target density of 120 plants m -2, as profitability increased and weed biomass decreased. Although profitable, higher seeding rates may be constrained by the increased cost of additional seed.

20 Table Days after planting (DAP) for plot management of organic field pea site-years Operation Vonda Delisle Vonda Vanscoy Seeding date 11 May 20 May 18 May 12 May In-crop harrow Tillage z Green manure ploughdown Tillage y Physiological maturity Harvest z Summerfallow treatment only y Summerfallow and green manure treatment Table Transformations used for organicallygrown field pea to satisfy the assumption of homogeneity of variance for PROC MIXED in SAS. Datasets combined from four site-years of data Variable Transformation Emergence Ln z Crop biomass y Weed biomass Ln Yield Harvest index None Hundred seed weight None Potential return Seed P concentration None Straw P concentration None Straw P uptake Seed P uptake z Natural log transformation y Square root transformation

21 Table Climatic data for sites in central Saskatchewan for seeding rate studies of organically-produced field pea Precipitation (mm) Month 30 yr average z y Vonda x Delisle w Vonda Vanscoy May June July August Total Mean daily temperature ( C) 30 yr average Vonda Delisle Vonda Vanscoy May June July August z Based on Saskatoon weather data from from Environment Canada y 2006 precipitation data gathered from nearest Environment Canada weather stations x Vonda 2005 climate data from nearest Environment Canada weather station (Osler, SK.) w Delisle 2005 climate data from nearest Environment Canada weather station (Saskatoon, SK.) Table Parameter estimates (±S.E.) for organic field pea crop and weed biomass and grain yield non-linear regressions based on Equations 4 and 5. Data are combined from four site-years of data Parameter estimates Variable Y max D 50 R 2 Crop biomass 6326 (351) z 110 (14) 0.99 Weed biomass 3101 (99) 92 (10) 0.99 Grain yield 2353 (61) 92 (6) 0.99 z ± standard error of the means in parentheses

22 Table Hundred seed weight and harvest index for organically-grown field pea combined from four site-years of data Seeding rate 100 Seed wt Harvest index viable seeds m -2 g 100 seeds -1 (seed:seed+straw) Seeding rate linear z (P>F) Seeding rate quad. (P>F) z Least squares means for linear and quadratic contrasts, significant at P 0.05

23 Table Seed and straw P uptake and concentration after harvest for organically grown field pea combined from four site-years of data Seeding rate Seed P conc. Straw P conc. Seed P uptake Straw P uptake viable seeds m -2 mg kg -1 - kg ha -1 kg ha -1 mg kg Seeding rate linear z (P>F) < < Seeding rate quad. (P>F) z Least squares means for linear and quadratic contrasts, significant at P 0.05

24 Table Soil nutrient availability after harvest for organic field pea combined from four site-years of data Seeding rate Available P Inorganic N viable seeds m -2 µg g -1 µg g -1 Delisle cm 0-15cm 15-30cm Summerfallow Green manure Vonda 2005 Summerfallow Green manure Vanscoy 2006 Summerfallow Green manure Vonda 2006 Summerfallow Green manure

25 Fig Stand establishment for pea. Points represent the mean of four site-years. Bars indicate standard error of the means. Fig Effect of seeding rate on pea crop and weed biomass at physiological maturity. Points represent four site years. Bars indicate standard error of the means. 25

26 Fig Effect of seeding rate on grain yield of organically grown pea. Points represent combined data for four site years. Bars indicate standard error of the means. Fig Return for organic field pea combined for four site-years. Bars indicate standard error of the means. Regression lines indicate average price received ( ), 2005 selling price (---), and 2006 selling price ( ). 26

27 3.2 Experiment 2 Optimal seeding rate for organic production of lentil in the northern Great Plains Introduction The vast majority of Canadian-grown organic lentil (Lens culinaris) (>99%) are produced in Saskatchewan. More than 36,000 ha were devoted to organic lentil production in 2005 (Macey 2006). While there are benefits to growing lentil in organic systems such as atmospheric nitrogen (N) fixation, this crop is poorly competitive with weeds due to its small stature and slow growth early in the season (Ball et al. 1997; McDonald et al. 2007). Lentil yield losses from weeds may be as high as 80% (Boerboom and Young 1995; Paolini et al. 2003). Increasing the seeding rate of lentil in organic systems may increase its competitive ability and decrease weed biomass. The recommended seeding rate for conventional production of lentil in western Canada is 130 plants m -2 (Saskatchewan Pulse Growers 2000). An unpublished study conducted at a single site in Saskatchewan determined that the optimal seeding rate for organic production of lentil is 195 to 260 plants m -2 and that row spacing should be narrow to allow for better crop competition with weeds for resources (Johnson 2002). European studies have determined that optimal seeding rates should exceed 130 plants m -2 where pesticides are not used, although many of the experiments were not conducted as organic production systems. In an Italian study, the optimal plant density for lentil when using mechanical weed control was 177 to 250 plants m -2 (Paolini et al. 2003). Similarly, Ball et al. (1997) found that increasing small-seeded lentil seeding rates had a suppressive effect on weed dry weight in the eastern United States. In an Australian field study, 27

28 lentil showed a strong response to seeding rate; increased rates increased lentil yield and decreased weed biomass up to 200 plants m -2 (McDonald et al. 2007). Another Australian study suggested a seeding rate of 150 plants m -2, but up to 230 plants m -2 where growing conditions were less favourable (Siddique et al. 1998). In a Turkish study, lentil was seeded at a rate of 350 plants m -2, which produced a plant density of 200 to 226 plants m -2. Yield loss due to weed interference was approximately 50% as compared to a hand-weeded check (Elkoca et al. 2005). A common theme throughout these seeding rate studies is that seeding rates should exceed the recommended rate for conventional production in western Canada. The optimal seeding rate may be dependent on the cost of seed; higher costs for organic seed may be prohibitive to increasing seeding rates beyond a certain level. In addition, the profitability of organic crops is dependent upon price premiums which can be highly variable over time (Smith et al. 2004). For example, the University of Saskatchewan (2006) reported that the price paid for organic lentil ranged from $0.88 to $1.98 per kg in While increasing seeding rate may increase the competitive ability of the crop, the profitability of the crop may or may not increase, as the cost and selling price of the seed must be taken into consideration (Siddique et al. 1998). The economic optimum seeding rate takes into account the cost and selling price of seed. Previous analyses have used various methods to determine economic optimum seeding rate. The most common method was developed by French et al. (1994), where a non-linear regression was fitted to describe the yield-density response, and then the slope where the optimal return was reached was determined and the ideal density identified (Loss et al. 1998; Jettner et al. 1999). Non-linear regression equations have also been fit to gross return (minus seed cost) and density 28

29 where the economic optimum plant densities were determined from the density where return was maximized (Norsworthy and Oliver 2001; Shirtliffe and Johnston 2002). While increasing seeding rates in many instances increases profitability, yield and competitiveness of the crop (Mohler 2001) the effects on soil fertility, plant nutrition and seed quality and are varied or unknown. The objective of this study was to determine an optimal seeding rate for organic production of lentil considering a number of factors, including yield, weed suppression, soil N and P concentrations, plant P uptake and profitability MATERIALS AND METHODS Experimental Design and Plot Management Sites were established on pre-existing organic farms in 2005 and The site locations were south of Vonda and west of Delisle, SK in 2005 and Vanscoy and Vonda, SK in Sites were organically managed for approximately 3 yr at Vanscoy, 8 yr at Delisle and 20 yr at Vonda. Site descriptions are given in Table Prior to plot establishment, sites were seeded to barley (Hordeum vulgare) in 2005 and to wheat (Triticum aestivum) in Climatic conditions during the growing season including precipitation and temperature were recorded by weather stations at the research plots. In each location lentil (Lens culinaris cv. CDC Sovereign) was seeded in a randomized complete block design with four replicate blocks that included five target densities of 15, 38, 94, 235 and 375 plants m -2 and two fallow treatments (green manure and summerfallow). The fallow treatments were included to compare soil fertility relative to seeding rate treatments. Indianhead lentil was used as the green manure crop and seeded to achieve a target density of 235 plants m -2 (Lawley 2004). In 2006 an additional green manure treatment was seeded at a density of

30 viable seeds m -2 to achieve the target density. The single plot size for each treatment was 2m x 6m. Lentils were seeded at a 23 cm row spacing at a depth of 2.5 cm using a cone seeder with an offset disc drill. The number of seeds planted was increased based on the percentage germination determined from a germination test to achieve target plant densities. When seeding, Nodulator granular Rhizobium inoculant (Becker Underwood, Saskatoon, SK) was placed with the seed at the recommended rate for lentil. Pre-seeding tillage was performed at all sites to manage early-emerging weeds. In-crop harrowing was performed with two passes using a tine harrow for seeding rate treatments when the crop was judged to be sufficiently resilient (approximately one month after seeding). Tillage with performed in the fallow treatments (two passes using a tandem disc) to incorporate the green manure treatment and to manage weeds as required. The natural weed population was used as it was relatively homogeneous within sites. Actual crop and weed species densities were determined by counting the number of each species in two randomly selected 0.25 m 2 quadrats in each plot approximately one week after the in-crop harrowing was performed. Aerial biomass sampling of both the crop and weeds occurred at physiological maturity (indicated by tan-coloured pods) and combined for each plot. Each sample was 0.25 m 2 and was cut one cm above ground level. Plant material was separated into crop and weed species. All plant material was dried at 60 C for 72 h and then weighed to determine biomass. A hand harvest of four 1-m long rows (equivalent to 0.81 m 2 ) was taken from each plot to determine seed yield and harvest index. Further information regarding management operations are given in Table

31 Soil analysis Soil samples for general site characteristics were taken at 0- to 15-cm and 15- to 30-cm depths at five random locations within each trial area just prior to seeding. Samples from each depth were combined and sent to ALS Laboratory Group (Saskatoon, SK) for ph, electrical conductivity (EC), and macronutrient analysis (Table 3.2.1). Two soil samples were taken after harvest in each plot in randomly selected locations at 0- to 15-cm and 15- to 30-cm depths and samples at each depth were combined within each plot. Each sample was air-dried and ground to pass through a 2 mm sieve. Subsamples of the ground soils were analyzed for determination of available soil nutrients. Available phosphate-p for the 0- to 15-cm depth was assessed by the modified Kelowna extraction method (Qian et al. 1994). Inorganic N (NH NO - 3 ) was assessed by KCL extraction for both the 0- to 15-cm and 15- to 30-cm depths (Maynard and Kalra 1993) by ALS Laboratory Group (Saskatoon, SK). Plant analysis Aerial biomass samples taken at harvest were air-dried in a covered facility in cloth bags for one week, then moved indoors and dried further at approximately 30 C. Samples were stored indoors until threshing. Samples were weighed prior to threshing and then threshed by machine. Seed weight was determined from the threshed, cleaned seed. Vegetative biomass weight was determined by subtracting seed weight from the total harvest biomass sample weight. Harvest index was determined from the ratio of the weight of seed:seed + straw. Seed quality was assessed using the Canadian Grain Commission s seed grading guides for lentil (CGC 2005). A further assessment of seed quality using 100 seed weight was also performed. Two samples of 100 seeds were weighed and averaged for each plot. 31

32 Incidence of disease was a concern for denser stands, so grain samples from the highest seeding rate were tested for seed-borne disease. Fifty seeds from each replicate were surface sterilized with a 10% bleach solution for two min, then plated using sterile technique onto potato-dextrose agar (10 seeds per Petri dish) and set onto lighted benches for one week (Morrall and Beauchamp 1988). Assessments of occurrence of disease were made by a plant pathology technician based on the colour and morphology of the colonies observed. After seed quality and disease incidence were assessed, seed and straw samples were ground for nutrient analysis. All samples were assessed for P content by acid digestion (Thomas et al. 1967). Phosphorus levels in the digested samples were determined colorimetrically using a Technicon AutoAnalyzer II (Technicon Industrial Systems, Tarrytown, N.J.). Statistical analysis Yield data were analyzed using the PROC MIXED procedure of SAS (SAS Institute Inc., 2004) with block and site-year as random effects and seeding rate as the fixed effect. Treatment effects were considered significant at P Data were tested and transformed as required to meet the assumption of homogeneity of variance. Back-transformed data are presented. Sites and years were combined as there were no significant interactions between site-year and treatment. Significant results are presented in-text with the F value, numerator and denominator degrees of freedom and P value enclosed by brackets. The Proc GLM procedure of SAS (SAS Institute Inc. 2004) was used to perform linear and quadratic contrasts to determine significant trends in data (other than yield data) and to compare means between summerfallow, green manure and the highest seeding rate treatments. 32

33 A non-linear regression model was fitted to the crop biomass and grain yield means using the PROC NLIN procedure of SAS (SAS Institute Inc. 2004). The two-parameter Michaelis- Menten model was used with a modification of parameters: [1] where Y max (kg ha -1 ) is the maximum yield as the density approaches infinity, D 50 (viable seeds m -2 ) is the crop density at which half of Y max occurs, and D (viable seeds m -2 ) is seeding rate. The D 50 parameter describes the slope of the curve. This model was chosen to describe these data because it provided a good fit and biologically meaningful parameters. The relationship between weed biomass and crop density was described using a modified version of the Michaelis-Menten equation to fit decreasing biomass: [2] The fit of the equation to the means was determined using the adjusted R 2 value in SigmaPlot 10.0 (Systat Software, Inc.) A non-linear regression model was fitted to the means of the return data combined for sites using the PROC NLIN procedure of SAS (SAS Institute Inc. 2004). The two-parameter Michaelis-Menten model was used: [3] where R max is the maximum return ($ ha -1 ), D 50 is the crop density at which half of R max occurs (viable seeds m -2 ), and D is the seeding rate (viable seeds m -2 ). The Michaelis-Menten equation was fitted to means resulting from the mixed-model ANOVA. 33

34 Economic analysis To determine the profitability of each seeding rate, an equation was used to calculate the potential return (Norsworthy and Oliver 2001): [4] where R is the return ($ ha -1 ), Y is the seed yield (kg ha -1 ), PR is the price received, SW is the seed weight planted (kg ha -1 ) and C is the seed cost ($ kg -1 ). The cost of seed was provided by an organic seed supplier near Saskatoon, SK (Marysburg Organic Producers Inc. personal communication). The seed cost used was $0.84 kg - 1. The seed weight planted was determined from the 1000 seed weight as counted and measured, and multiplied by the number of seeds planted for each seeding rate. The number of seeds planted depended upon the percent germination, as described above. The gross profit was determined from the mean selling price in Saskatchewan averaged for 2005 and 2006 as reported by the University of Saskatchewan (2006). The mean selling prices for lentil in 2005 and 2006 were $1.32 and $0.77 kg -1, respectively. The average selling price used was $1.05 kg -1. Three hypothetical scenarios were also assessed based on seed cost being equal to selling price, seed cost being 1.5 x selling price and 2 x the selling price to determine potential effects of fluctuations in seed cost on return and economic optimum seeding rate RESULTS AND DISCUSSION Weather Climatic conditions varied between years and from the 30 y average for the Saskatoon area (Table 3.2.3). 34

35 Crop Emergence and Growth Low establishment rates resulted in lower crop densities than expected. Stand establishment increased linearly as seeding rate increased (slope = 0.50, R 2 = 0.99) (Fig ). Establishment rates varied from 67% at the lowest seeding rate to 49% at the highest seeding rate with no trend as seeding rate increased. Crop biomass at physiological maturity increased asymptotically as seeding rates increased (Fig ). Biomass increased from 314 to 3998 kg ha -1 between 15 and 375 viable seeds m -2. The increase in biomass was significantly related to seeding rate (F198.98, 5 = 33.8, P<0.0001). One-half of the predicted maximum biomass accumulation occurred at 151 viable seeds m -2 (Table 3.2.4). Weed biomass decreased by 59% (from 3155 kg ha -1 to 1298 kg ha -1 ) as seeding rate increased (Fig ). Seeding rate effects were significant for weed biomass reduction (F371.30, 5 = 13.6, P <0.0001). Weed biomass was reduced by 26% above the rate recommended for conventional production of lentil of 130 plants m -2. These results are consistent with the findings of other studies. An increase in lentil yield decreased canola yields by approximately 25% between 80 and 200 plants m -2 where canola was considered a weedy species (McDonald et al. 2007). Ball et al. (1997) found that weed dry weight was reduced by 20 to 60% when the seeding rate of lentil doubled from 63 to 126 plants m -2, depending on seed placement. Weed biomass also decreased as lentil density increased to 359 plants m -2 in a study by Paolini et al. (2003). Grain Yield Lentil seed yield was significantly linearly related to seeding rate (F98.84, 5 = 33.8, P<0.0001) (Table 3.2.4). Seed yield increased with increasing seeding rate (Fig ) from