Low Input Weed Management in Field Peas

Two trials were conducted on a Templeton silt loam soil o 28' E.) in 2007/08. The aim was to compare the competitive ability of different pea canopy architectures as influenced by genotype, population, sowing date and their interaction as a means of low input weed control strategy. The first experiment had three sowing dates, two pea genotypes and two herbicide treatments. Experiment 2 treatments were a factorial combination of four pea populations and three sown artificial weed populations. A significant sowing date x pea genotype interaction showed that in the August sowing genotype had no effect on seed yield. However, in September sown plots Pro 7035 yielded 559 g m-2 , which was 40% more than Midichi, and in the October sowing, the difference was 87% more. Herbicide-sprayed peas produced 19% more seed (508 g m-2) than the unsprayed plants. When no weeds were sown, the highest pea total dry matter (TDM) of 1,129 g m-2 occurred at 200 plants m-2. This was more than twice (513 g m-2) the yield of the lowest population (50 plants m-2). There was distinct variation in the weed spectrum over time. Coronopus didymus, Stellaria media and Lolium spp were present in relatively large numbers throughout the season. Some weeds only occurred late in the season meaning they could be successfully controlled by early sowing. It could be concluded that it is possible to obtain high pea yields by using the right sowing date and appropriate seed rate as a means of low input weed management strategy.


INTRODUCTION
The poor ability of pea crops to compete with weeds [1,2] is the major drawback of growing them under low input or organic systems.Weeds can cause severe yield losses if crops are not monitored closely, particularly during the early stages of weed emergence [3].Generally, poor weed management results in weed accumulation and a larger weed seed bank.Farmers usually use conventional herbicides to manage weeds but low input farmers try to use less of that and the use of synthetic herbicides is not allowed in organic production systems.They currently mostly rely on cultural control methods.Weed control is therefore a real constraint in these systems.Some methods to control weeds under low input systems include intercropping and crop rotation [4], use of competitive crop genotypes [5,7], mechanical and hand weeding, use of appropriate sowing date and, often, high sowing rates [8].Several crops show genotypic differences in their competitive ability [8,9] mostly related to plant architecture, leaf area, leaf angle, plant stature, seed and seedling vigour.Also different weed species have different competitive abilities with crops [10].
Viability of low input and organic systems depends on achieving acceptable yields.Freeman [3] stressed that consistent yields of around 4 t ha -1 are necessary for field *Address correspondence to this author at the Department of Primary Industries, Victoria, Australia; Tel: 0061409511032; Fax: 0061-3-54304304; Emails: munakamwez@yahoo.com,zachariah.munakamwe@dpi.vic.gov.aupeas to be a viable crop.According to Moot [11], White and Hill [12], these high pea yields are achievable under favourable conditions despite peas' poor yield stability [13,14].
The research objective of this work was to compare the competitive ability of different pea canopy architectures as influenced by genotype (conventional and semi-leafless), population, sowing date and their interaction as a means of low input weed control strategy.

MATERIALS AND METHODS
Two trials were conducted in 2007/08 on a Templeton silt loam soil [15] at the Horticulture Research Area, Lincoln University, Canterbury, New Zealand (43° 38'S, 172° 28' E.).MAF soil quick tests [16] were done to establish actual soil available nutrient levels (Table 1).All the nutrient levels were in the acceptable range for growing peas and the pH was also optimal.
In experiment 1, treatments were arranged in a split plot design with three replicates.Main plots were sown on 9 August, 13 September and 15 October 2007.Sub-plots were a factorial combination of two pea genotypes, conventional (Pro 7035) and semi-leafless (Midichi) and two herbicide treatments (cyanazine at 0 and 500 g a.i.ha -1 ) applied before emergence.The total number of plots was 54 (36 plots with peas and 18 no pea control plots).Each plot was 2.1 m wide x 10 m long.Experiment 2 was sown on 13 September and the treatments were a factorial combination of four pea populations (0, 0.5 x recommended sowing rate, recommended sowing rate (100 plants m -2 ), 2.0 x recommended), and three sown artificial weed populations (0, 1/3 recommended (referred to here as lower rate) and 2/3 recommended (referred to here as higher rate) of each weed.The sown artificial weeds were a mixture of Brassica napus, Lolium multiflorum and Vicia sativa which had recommended sowing rates of 3, 25 and 30 kg ha -1 respectively when sown as crops and this translated to 100, 833 and 75 seeds m -2 respectively.This was a good representation of a broad spectrum of weeds commonly found in most fields.The experiment design was a randomised complete block with three replicates.The total number of plots was 36.Each plot was 2.1 m x 6 m long.The field pea variety used was Midichi (a semi-leafless type).

HUSBANDRY
Land was prepared using conventional methods, i.e. disking, rolling and harrowing.It was tilled to a depth of 25 cm.Seed was drilled with an Öyjord cone seeder at a depth of 5 cm.In experiment 1, seed was sown in 15 cm rows and was sown at 100 plants m -2 at the above-stated sowing dates.Wakil, a formulated mixture of metalaxyl, fludioxonil and cymoxanil for the control of Peronospora spp (downy mildew), Pythium spp and Ascochyta spp, was applied to all seed at the equivalent of 2 kg t -1 of seed before sowing.All sowing rates were corrected for germination percentage and expected field emergence for each pea variety.Experiment 2 was sown on 13 September in 15 cm rows with varying intra row spacing to achieve pea populations of 50, 100 and 200 plants m -2 .The sown weed seed was then broadcasted onto plots and lightly harrowed to incorporate them into the soil.
Irrigation was applied based on crop requirement as determined by Time Domain Reflectometry (TDR) in the 0 -20 cm soil layer, when the soil reached 50% of field capacity based on the first sowing date.A mini boom irrigator applied 30 mm of water at each irrigation.A total of 120 mm was applied to both experiments.The peas were sprayed with cyproconazole at 250 ml ha -1 to combat powdery mildew (Erysiphe spp) and with copper oxychloride at 1 kg ha -1 for downy mildew in both experiments.

MEASUREMENTS AND ANALYSIS
A 0.2 m 2 sample was taken from each plot using a 0.1 m 2 quadrat every 7-10 days throughout the season starting from three weeks after crop emergence.This was used for crop and weed dry matter measurements.Samples were dried in a forced draught oven for 24 -48 h at 60 °C to a constant weight and then weighed.Final harvests were taken when crops reached a moisture content of 15 -18%.Final total DM and seed yield were estimated from 1 m 2 quadrat samples.Plants were cut at ground level and weighed.They were hand threshed and the seeds weighed.Weed counts were taken three times during the growing season and this was at 10 weeks after emergence of each sowing date.Weeds were sorted by taxa (species or genus depending on similarity) and counted.Uncommon taxa were pooled and their total count recorded.
All data were subjected to the analysis of variance (ANOVA).Genstat 10.1.Copyright 2007, Lawes Agricultural Trust (Rothamsted Experimental Station) was used for statistical analysis.Means were separated at the 5% level of significance using least significance difference (LSD) for sowing date, herbicide, genotype, population and interactions effects.

Climate
Climate data were from the Broadfields Meteorological Station, Lincoln University located about 1.5 km from the experimental site.The 2007/08 growing season was generally dry, with January rainfall being just 38% of the longterm average (Fig. 1).Substantial rain fell at the end of the season in February (104 mm).The season was generally cool and all mean temperatures, except in September, were lower than long-term means (Fig. 2).

Crop Yield and Harvest Index
Total DM at final harvest of the August and September sowings were not significantly different (mean 1,018 g m -2 ) but they were significantly higher was from the October sowing and cyanazine sprayed plots produced 21% more TDM than unsprayed plots (788 g m -2 ) (Table 2).There was no significant difference in the mean TDM produced by the two pea cultivars Midichi and Pro 7035 (mean 941 g m -2 ).
Herbicide sprayed peas produced 19% more seed (508 g m -2 ) than the unsprayed plants (Table 2).A significant (p < 0.05) sowing date x pea genotype interaction showed that in the August, sowing genotype had no effect on seed yield (Table 3).However, in September sown plots Pro 7035 yielded 559 g m -2 , which was 40% more than Midichi, and in the October sowing, the difference was 87% more.
Herbicide had no effect on crop harvest index (CHI).Pro 7035 had a higher CHI than Midichi (0.56).There was a significant sowing date x genotype interaction for CHI (Table 4).This showed that in August sowing there was less difference in CHI between the two cultivars than at the other two sowing dates.
In experiment 2, dry matter accumulation was directly proportional to pea population throughout the season and growth curves for each population had a typical sigmoidal shape (Fig. 3).The highest pea TDM was achieved at 200 plants m -2 (1,120 g m -2 ), which was more than twice the yield        of the lowest pea population (513 g m -2 ) with sown weeds (Table 5).The control treatment (no-sown weeds) had the highest pea DM throughout the season.The low weed rate and the high weed rate treatments had similar DM accumulation throughout.However, the two were significantly different from the control treatment (Fig. 4).In experiment 2 seed yield increased significantly (p < 0.001) as pea population increased (Table 5).Two hundred pea plants m -2 gave the highest mean seed yield at 409 g m -2 and 50 pea plants m -2 the lowest at 197 g m -2 .On the other hand the control treatment gave the highest mean seed yield of 390 g m -2 .CHI did not vary and the grand mean was 0.39.

Total Weed Dry Matter
In experiment 1 there was no difference in weed DM accumulation in response to pea genotype throughout until harvest when the no pea treatment plots had the highest weed DM (Fig. 5).Throughout the season there was more weed Days after emergence DM in unsprayed plots than in sprayed plots (Fig. 6).In experiment 2, weed DM always increased with decreased pea population throughout the season (Fig. 7).At final harvest, there was a 31% reduction in weed DM with an increase in pea population from 0 to 50 plants m -2 and a similar percentage decrease from 50 to 100 plants m -2 (Table 6).
Overall, there was a 51% reduction in weed dry matter from 50 to 200 plants m -2 .With sown weeds there was an increase in weed DM with increased weed population.The no-sownweed control plots had the lowest weed biomass throughout the season (Fig. 8).However, weed DM in the two sown weed treatments were not significantly different from each other but were significantly different from the no-sown weed treatment throughout the season.

Weed Counts
There was distinct variation in the weed spectrum over time in experiment 1. Tables 7, 8 and 9 show weed counts for each sowing date.Generally, weed counts were lower in sprayed than in unsprayed plots and there were several sig      nificant herbicide x pea genotype interactions on most major weeds.To summarise the interactions, significant differences of weed counts between the cyanazine sprayed plots and unsprayed plots was highest in the no pea control plots, followed by Midichi plots and the lowest was in Pro 7035.

DISCUSSION
A significant (p < 0.05) sowing date x genotype interaction showed that in the August sowing genotype had no effect on seed yield.However in September sown plots the Pro 7035 seed yield of 559 g m -2 was 40% more than that Midichi.By October it was 87% more.This highlights the need to select a suitable genotype to use at different times in the season.Early in the season both genotypes could be used without yield reduction but as the season progressed, it was better to use a fully leafed genotype to smother the increased weed spectrum and numbers associated with the later sowing date, although both pea types were significantly better than the control no pea plots.
Genotype had no effect on seed yield in August because there were fewer weeds, which were slow growing with the low temperatures.This gave both pea genotypes (base temperature 4 °C) the same competitive advantage over the weeds and hence the effect of weeds was not evident in that sowing.However, there was an increase in weed spectrum and quantity as the season progressed possibly attributable to increased temperatures so the effect of weeds and the differences in pea competitive ability against them of the different genotypes became evident.
Herbicide was effective in reducing weeds.Sprayed plots had a mean seed yield of 508 g m -2 , which was 19% more than the mean of unsprayed plots.This shows the effect of weeds on crop yield through competition for nutrients, light, space, and water.
Seed yield increased significantly (p < 0.001) as pea population increased.At 200 plants m -2 the highest mean seed yield of 409 g m -2 was obtained and at 50 plants m -2 it was the lowest (197 g m -2 ).Similarly, Townley-Smith and Wright [17] reported pea yield increases and weed DW reduction by raising field pea density from 50 to 100 seeds m -2 , but concluded that increasing the seeding rate over 100 seeds m -2 would be unlikely to give a better result.According to them, a 70% increase in the seeding rate (150 seeds m -2 compared with normal 90 seeds m -2 ) was costly in peas and could not always be compensated for by higher yield.Martin et al. [18] reported that increased plant density above 150 plants m -2 was not associated with a higher seed yield, although it did increase straw production.Similarly, White and Hill [12] recommended an optimum population of 70 plants m -2 on shallow soils, 90 plants m -2 on deeper soils and 100-120 plants m -2 for irrigated pea crops in New Zealand.McKenzie et al. [19] reported optimum dry pea populations of 90 -100 plants m -2 but did not specify growing conditions.
Weed DM production was inversely proportional to pea population from 42 DAE until final harvest (Fig. 7).Increased pea population gave the crop a greater competitive advantage against weeds and a relatively higher TDM production and seed yield.The no-sown artificial weed treatment gave the highest mean seed yield of 390 g m -2 because it had just few weeds hence experienced the least competition.The reduction in pea TDM with increased weeds was basically because of competition for light and nutrients.Peas can clearly out compete weeds for light if sown at a higher than normally recommended population [8].[20] reported that higher seeding rates of peas are effective in reducing weed development and Farshatov [21] found that raising the sowing rates of peas from 100 -140 plants m -2 reduced the weed population 2.5 fold.In this experiment there was a 31% reduction in weed DM with increased pea population from 0 to 50 plants m -2  and a similar percentage reduction from 50 to 100 plants m -2 .Overall there was a 51% reduction from 50 to 200 pea plants m -2 .Grevsen, [22] found a similar weed reduction and reported that increasing the seeding rate from the normal 90

Table 4 . The Sowing Date x Pea Genotype Interaction on CHI of Field Peas Grown in Canterbury in the 2007/08 Growing Season (Experiment 1)
**p<0.01