Introduction
The lack of effective weed management practices is widely reported to be a major constraint in organic productionReference Walz1. In the absence of widely supported production strategies and extension programs usually available in conventional agriculture, organic farmers have devised organic weed management (OWM) solutions based on their own needs and capabilities with support from grower groups and certification organizations, and information from published materials in Australia and overseas. But the effectiveness of many of these systems has not been examined.
In Australia, only a few farming systems trials have been conducted that evaluated the effect of organic and non-organic production systems on crop yield, weed density, soil fertility, costs and other variablesReference Penfold, Miyan, Reeves and Grierson2, Reference Wells3. These trials have not studied the comparative effects of different weeding methods within a given farming system. A small amount of research has been conducted on specific non-chemical weeding methods in herb or vegetable systems, including mulchesReference Olsen and Gounder4 and tillageReference Dunn, Penfold, Kristensen and Hogh-Jensen5, but the economics of weed management methods in annual horticultural cropping in Australia has received little attentionReference Buntain6. Several overseas studies compared different weeding methods in organic vegetable production, including some that present an economic evaluation of the methods testedReference Melander, Foguelman and Lockeretz7. Factors that are considered to have an impact on the relative economic performance of weed control methods include the existing weed density, size of the area to be treated, crop type, cost of labor and the price received for the crop.
This paper reports on a comparative evaluation of the agronomic and economic performance of several weed control methods frequently used in organic herb and vegetable production in Australia. In a survey of Australian organic herb and vegetable growers, hand weeding (e.g. hand hoeing) was the most common method used (95%), followed by mulching (78%), tillage before and during the cropping season (70%), slashing or mowing to reduce seed set (69%) and crop rotations (65%)Reference Kristiansen, Jessop, Sindel, Rowe, Mendham and Donaghy8. The methods evaluated in the trials reported here (hand weeding, tillage and organic mulches) are also frequently reported in other surveys of organic growersReference Walz1. In addition to hay mulch, a novel pelletized paper mulch (PP) was used in the experiments. PPs have been reported to be effective for weed suppressionReference Smith, Gilliam, Edwards, Olive, Eakes and Williams9.
A series of experiments was conducted from 1999 to 2001 on two crops with contrasting growth habits, crop growth durations and wholesale values. The medicinal herb, echinacea (Echinacea purpurea Moench. [L.]), is widely grown by organic herb growers in AustraliaReference Kristiansen, Jessop, Sindel, Rowe, Mendham and Donaghy8. The herb has a relatively high economic value but production is hindered by a long growing season in which weed control is reported to be a serious problemReference Buntain6. The salad vegetable, lettuce (Lactuca sativa L. cv. Imperial Triumph), in contrast, has a short growing season, forms a smaller canopy and is a less valuable crop. Effective weed control is reported to be more straightforwardReference Roberts10. The experiments were designed to assess the relative performance of various weed control treatments in terms of weed growth, growth of contrasting crops and treatment cost.
Methods
Trial sites
The field experiments were carried out at three sites over three consecutive summer growing seasons on the Northern Tablelands of New South Wales. In the first year (1998–1999) field trials were conducted at Yarrowitch (31.14°S, 152.00°E, elevation 970 m). This property had been operated organically for the previous two seasons, before which it was used for conventional wool growing. The soil was a granite-derived fine sandy loam (yellow chromosolReference Isbell11) with 2% organic carbon and pH 6.0.
In the second year (1999–2000) field trials were undertaken on two of the University of New England's research farms: Kirby Research Station (30.43°S, 151.61°E, elevation 1135 m), and Laureldale Research Station (30.48°S, 151.65°E, elevation 1063 m). Kirby is principally a wool-growing property and the paddock in which the trials were conducted had been sown to pasture and grazed by sheep and cattle for at least the previous 7 years. No herbicides or fertilizers had been used during that time. Laureldale is regularly used for small-plot field trials and has a complex land-use history based on a range of crops and a diverse mixture of fertilizer and pesticide usage patterns. The soil at Kirby was a granite-derived sandy loam (yellow chromosolReference Isbell11) with 1.7% organic carbon and pH 5.3, while the Laureldale soil was a basalt-derived chocolate clay (brown dermosolReference Isbell11) with 1.8% organic carbon and pH 5.4. In the third year (2000–2001) field trials were not continued at Laureldale due to poor crop growth in the second season, heterogeneity of the site (e.g. variability of weeds), and the high clay content of the soil making access difficult after rain.
The Yarrowitch site was chosen initially as it was a certified working organic farm. When that site became unavailable due to management changes, the trials were conducted on two research farms in the Armidale area with similar climatic conditions but contrasting soil types that are very common in the region. Yarrowitch has a slightly warmer climate with higher rainfall compared with the sites near Armidale, but the soil is similar to Kirby. A summary of the climatic data in Yarrowitch and Armidale during the trials is shown in Figure 1. In the lettuce trials, the rainfall at the Armidale sites in 2000 was lower than Yarrowitch in 1999 and Kirby in 2001. The rainfall at Yarrowitch in 1999 and Kirby in 2001 was similar for the lettuce trials but not the echinacea trials, where rainfall was greater at Yarrowitch compared with Kirby (700 and 400 mm cumulative rainfall respectively).
Monthly summaries of average maximum and minimum temperatures and rainfall at Yarrowitch and Armidale for the periods in which the field trials were conducted.
Experimental design and treatments
Five weed control treatments were assessed, with echinacea and lettuce used as the test crops. The echinacea germplasm was sourced from an organic grower in the region and the cultivar is unknown. Lettuce seed was purchased from a commercial supplier and Imperial Triumph, a crisp head cultivar, was used. Four replicates for each treatment (except lettuce at Kirby, where eight replicates were used) were laid out in a completely randomized design. A control treatment (CO), in which weeds were allowed to grow unchecked, was used to determine the background weed load and the comparative effect on crop yield of doing nothing to control weeds. A hand weeding treatment (HD) was used to test its effectiveness in controlling weeds, and to act as a weeded pseudo-control for determining crop yield in a situation with minimal competition from weeds. Hand weeding was carried out at 4 weeks after planting (WAP) in lettuce and at 4, 8 and 12 WAP in echinacea, and consisted of manually removing weeds using a wheel-mounted stirrup hoe for weeding along the rows and a chipping hoe for removing weeds between crop plants in the planting row. The wheel hoe blade was 230 mm wide and was used at a depth of 50 mm, and the chipping hoe blade was 100 mm wide and used to a depth of 100 mm. The tillage treatment (TI), carried out at the same time as hand weeding, consisted of mechanically removing weeds using a tractor-mounted rigid tine chisel plough. The plough had seven 50 mm wide tines operating at a depth of 130 mm. After each tillage operation, a chipping hoe was used to remove weeds growing in the planting row between the crop plants but not those in close proximity to the crop plants. The hay mulch treatment (HY), predominantly ryegrass (Lolium L. sp.) and oats (Avena sativa L.), was applied to a thickness of 100 mm (9.5 tonnes ha−1) after planting, and at 4 WAP weeds immediately around the crop plants were manually removed by hand and a further 0.5 tonnes ha−1 of hay was added. The PP, composed of pelletized and dried waste paper slurry, was applied to a thickness of 30 mm (42 tons ha−1) after planting. At 4 WAP, weeds immediately around the crop plants were manually removed by hand. The PP was not used in the 1999–2000 trials due to lack of availability. The C:N ratio of the mulches was 39:1 and 171:1 for hay and paper respectively. A summary of the treatments used in each trial is presented in Table 1.
Summary of experiments and the treatments applied in each experiment.
The experimental plots were prepared using a chisel plough and rotary hoe with the rows being 2 m wide. Echinacea seedlings, aged approximately 8 weeks, were transplanted on October 20, 1998 (Yarrowitch) and November 15, 2000 (Kirby). The seedlings were planted into plots 7 m long in two rows 600 mm apart with a 300 mm spacing between plants along the rows. Lettuce seedlings, aged approximately 3 weeks were planted on December 11, 1998 (Yarrowitch), January 7, 2000 and January 8, 2001 (Kirby and Laureldale 2000). A similar configuration to the echinacea experiment was used but with plots 10 m long. The plants were fertilized by hand after planting with ‘Long Life’ Dynamic Lifter®, a pelletized poultry manure formulation (nitrogen 4%, phosphorus 3.1%, potassium 1%, calcium 7%, magnesium 0.3%, zinc 0.02 mg kg−1 and manganese 0.02%), at a rate of 70 g per plant (700 kg ha−1). Immediately after planting, the seedlings were watered-in using approximately 5 liters of water for each plant. No further irrigation was applied for either crop, which is typical practice for echinacea growers but not lettuce. The echinacea was harvested on March 22, 1999 and April 4, 2001 and the lettuce on February 5, 1999, February 26, 2000 and February 28, 2001.
Measurements
Relative cover, i.e., the proportion of ground area occupied by weeds or crops, was determined using photographs taken above the center of the plotsReference Lutman12, Reference Siddique, Belford, Perry and Tennant13. The area captured in the photos was 0.87 m2 (1.11 m long and 0.78 m wide), including the crop rows and a 250 mm wide strip outside the rows. The photographs were placed under a pane of glass with an etched grid of 28×20 cells (6.3 mm sides) and the cells counted according to the predominant ground cover category, either weed, crop, bare ground or mulch. Relative cover was assessed at 2, 4 and 7 WAP in the lettuce trials, and 4, 6, 8, 10, 12, 16, 18, 20 and 21 WAP in the echinacea trials.
Weeds were defined as any non-crop plant growing in the experimental plots. The predominant weed species at each site were recorded, but weeds were not separated by taxon or type. Weed density was measured at crop harvest using a 0.5 m×1 m quadrat placed lengthwise in the middle of the plot. Weed biomass was measured at harvest by placing a 0.5 m×1 m quadrat randomly within the plots, cutting all weeds at ground level, drying at 80°C for 72 h and weighing.
Echinacea biomass was determined at 21 WAP by collecting 10 whole plants from the center of the plot, roughly chopping, drying at 80°C for 72 h and weighing. Eight lettuce plants were harvested from the center of each plot at 7 WAP by cutting at ground level and weighing. Gross crop value (GCV) (AU$ ha−1) was calculated by multiplying crop biomass (kg ha−1) by the wholesale price ($ kg−1) The prevailing wholesale prices were $0.58 kg−1 for lettuceReference Heisswolf and Jackwitxz14 and $20.00 kg−1 for echinacea roots and $4.00 kg−1 for echinacea shoots (P. Green, personal communication).
A partial gross margin analysis was conducted to determine the cost effectiveness of weed control treatments. The economic analysis included costs due to the various treatments, with all other variable and fixed costs associated with the production of the crop assumed to be equal. The cost of each weed control treatment ($ ha−1) was based on the amount of materials used, the labor required and machinery usage. Adjusted crop values (ACVs) were derived by subtracting the cost of each treatment at each site from the GCV. The adjusted values provide a measure of the relative cost effectiveness of the treatments in each trial.
The cost for the HD treatment consisted of labor only, while the TI treatment costs included tractor/implement usage, driver's labor and supplementary chipping along the planting rows. The costs of the two mulching treatments were based on the price of the mulch, labor to apply the mulch and labor for supplementary hand weeding at 4 WAP. The costs for machinery and implement usage were obtained from the farm manager at Subiaco Herbs (P. Brown, personal communication) and at Kirby Research Station (N. Thomas, personal communication). There were no costs associated with the CO treatment. Labor was assumed to cost $15 h−1, including 20% on-costs and the mulches cost $110 ton−1 and $200 ton−1 for hay and paper respectively, not including transport. The amount of mulch required was 8 tons ha−1 of hay mulch and 42 tons ha−1 of PP. The cost of the TI treatment, including tractor, ploughing implement and the driver's labor, was $48.82 ha−1 at Yarrowitch and $56.76 ha−1 at Laureldale and Kirby.
To simplify comparisons between sites and years, time was expressed as degree days (°Cd). Degree day estimates were calculated using the single sine methodReference Roltsch, Zalom, Strawn, Strand and Pitcairn15, with a base temperature of 4.4°C used for lettuce and 5°C for echinaceaReference Ash, Blatta, Mitchell, Davies, Shaykewich, Wilson and Raddatz16.
Statistical analysis
Most analyses were carried out using generalized linear models in S-Plus 200017. Diagnostic checks were routinely made regarding distribution of the residuals and dispersion. Significantly different treatment means were separated by contrast analysis. Measurements repeated over a period of time were analyzed with mixed models using SAMMReference Butler, Gilmour, Cullis and Gogel18 in S-Plus. Time, trial, weed control treatment and their interactions were evaluated as fixed effects while time and plot location within the trials were used as random effects. The suitability of the models was assessed by inspecting the plot of residuals versus fitted values and by observing which random terms increased the log residual likelihood. In addition to linear regressions, non-linear regression was used for several analyzes. Relative weed cover for CO and TI in echinacea was fitted to the logistic function,
where y is the response variable, x is time (degree days), a represents the asymptote, b represents the inflexion point, and c is a scale parameter on the x-axis17. Relative echinacea cover was also fitted to Equation 1.
Results
Weed species observed
The weed flora at Yarrowitch was heavily dominated by the three grasses Digitaria sanguinalis (L.) Scop., Paspalum dilatatum Poir. and Setaria pumila (Poir.) Roem. & Schult., with almost no individuals of broad-leaved weeds achieving canopy dominance. At Laureldale, the weed flora was predominantly composed of the broadleaved weeds Polygonum aviculare L. and Hibiscus trionum L., and the grasses Echinochloa crus-galli (L.) P. Beauv. and D. sanguinalis. The trials at Kirby were mostly infested with Acetosella vulgaris Fourr., P. aviculare, Festuca arundinacea Schreb. and Conyza bonariensis (L.) Cronquist in both growing seasons (2000 and 2001). Vulpia myuros (L.) C.C. Gmel. became more common in the second season. The spatial distribution of weed species was highly variable and was not related to the treatments used in the experiment.
Weed growth in echinacea
The response to CO and TI were fitted to a logistic function (Eqn 1), while HD, HY and PP were fitted to a linear equation (Fig. 2). Rapid weed growth was observed in the unweeded CO with 90% cover achieved by about 6–8 WAP. Growth was limited in the TI plots to 40–60% maximum cover. For these treatments, the initial increase in relative weed cover was greater at Yarrowitch which was reflected by lower b (inflection point) values (b=286 and 553 for CO and TI respectively) than Kirby (b=448 and 2287). The HD, HY and PP treatments were effective in maintaining final weed cover levels of about 10% or less. The increase over time was significant for HD in both trials and PP at Kirby (P⩽0.002), but the final coverage was still less than 5%. The TI treatment was more variable than the other treatments, as indicated by the larger confidence limits in Figure 2.
Effect of treatments on relative weed cover in the echinacea trials. The circles show the data points, the solid lines are the regression curves, the shaded area represents the 95% confidence limits and the equations describe the regression for each treatment.
Weed density at echinacea harvest (Fig. 3A) was significantly different at the two sites (P<0.001), partly due to higher rainfall at Yarrowitch. The CO treatment had 930 weeds m−2 at Yarrowitch and 560 weeds m−2 at Kirby. The treatment effect on weed density was also highly significant (P<0.001), but the interaction between trial and treatment was not (P=0.72). Averaged across trials, the highest densities were recorded for CO (745 weeds m−2) and TI (401 weeds m−2), while HD (85 weeds m−2), HY (72 weeds m−2) and PP (38 weeds m−2) produced similar weed densities (P>0.68). Tillage reduced the density of weeds compared with CO by about 46%, whilst the HD and the mulches reduced weed densities by between 89 and 94%. Weed biomass at echinacea harvest (Fig. 3B) showed the same trends as for weed density. Weed biomass was greatest in CO (11,200 kg ha−1 at Yarrowitch and 6500 kg ha−1 at Kirby) and moderately high in the TI (average of 3500 kg ha−1 for both sites). The final weed biomass of the other treatments, ranging from 150 to 200 kg ha−1, did not differ significantly (P>0.88). A 60% reduction in weed biomass by TI compared with unweeded plots was observed, while HD, HY and PP reduced weed biomass by an average of 98%. Weed density and biomass at echinacea harvest were moderately correlated with relative weed cover across all trials (R 2=0.78 and 0.72 respectively).
Mean and standard error of weed density (A) and weed biomass (B) in echinacea for the weed control treatments. The data are presented on a log scale.
Weed growth in lettuce
In the lettuce trials at Yarrowitch in 1999, Laureldale and Kirby in 2000 and Kirby in 2001, relative weed cover varied significantly between trials and treatments, and the interaction was significant (P<0.001). The weed growth rates for CO at Yarrowitch and Kirby in 2001 (b≈0.11) were more than double those at Laureldale and Kirby in 2000 (b≈0.05). The dominant weeds at all sites were annuals, except A. vulgaris at Kirby.
Linear regression indicated statistically significant differences in slope; the greatest rate of weed growth occurred in CO, with moderate growth in TI and much lower growth rates in the other treatments (Fig. 4). Weed cover in HD did not increase over time, HY increased significantly in 2000 and 2001 and PP increased in 1999 only.
Effect of treatments on relative weed cover in the lettuce trials. The circles show the data points, the solid lines are the linear regression curves, the shaded area represents the 95% confidence limits and the equations describe the regression for each treatment. The PP treatment was not used in 2000.
Weed density and biomass at lettuce harvest (Fig. 5) also differed significantly between trials and treatments, and the interaction was significant (P<0.001), and the density and biomass measurements showed good correlation with relative weed cover (R 2=0.74 and 0.85 respectively). Density was highest at Yarrowitch (715 weeds m−2), more than double the density at Kirby in 2001 and about 10 times that at Laureldale and Kirby in 2000. Importantly, there was a large increase in weed density between the two seasons at Kirby; from 50 weeds m−2 (for CO) in 2000 to 335 weeds m−2 in 2001. The CO treatment usually had the highest weed densities, except at Laureldale where it was equivalent to TI and HY, while weed populations were smaller for HD. Averaging the treatments for each trial, the CO yielded 259 weeds m−2, TI 80 weeds m−2, HY 48 weeds m−2, PP 27 weeds m−2 and HD 14 weeds m−2. When weed loads were lighter (e.g., <100 weeds m−2 for CO), TI reduced weed density more than when weed pressure was high. This response was not observed for HD, HY and PP, which were more consistent between trials. Weed biomass at lettuce harvest followed a similar pattern to weed density, with CO producing the greatest weed biomass (1450 kg ha−1 averaged across trials) and HD the least (59 kg ha−1). Compared with CO, TI reduced weed biomass by 66%, HY by 85%, and HD and PP by 96%.
Mean and standard error of weed density (A) and weed biomass (B) in lettuce for the weed control treatments. The data are presented on a log scale. The PP treatment was not used in 2000.
Echinacea growth
The response of relative echinacea cover to the weed control treatments at Yarrowitch in 1999 and Kirby in 2001 was fitted to a logistic function (Eqn 1). There was very little increase in echinacea relative cover over time for the CO in both trials (Fig. 6). At Yarrowitch, HD and HY reached asymptotes of about 96%, PP reached 86% cover and TI achieved only 62% cover. Echinacea growth at Kirby was considerably lower than Yarrowitch because rainfall was lower by 43% in the second trial. Although echinacea is somewhat tolerant of dry conditionsReference Chapman and Auge19, commercial yields clearly require adequate irrigation. HD and HY had the highest crop cover (~65%), while TI and PP had significantly lower relative covers (~25%). An analysis of the effect of cumulative rainfall on relative crop cover showed that echinacea growth was closely related to rainfall (R 2⩾0.76), except in the weedy CO plots with very low crop cover.
Effect of treatments on relative echinacea cover. The circles show the data points, the solid lines are the logistic regression curves, the shaded area represents the 95% confidence limits and the equations describe the regression for each treatment.
Echinacea dry weight at harvest (raw data not shown, see Fig. 8A for derived values) showed similar responses as relative echinacea cover, with significant differences occurring between trials and treatments, and the interaction was significant (P⩽0.007). The yields for HD, TI and the mulched treatments at Yarrowitch were similar to commercial yieldsReference Buntain6. Correlation of biomass and relative cover yielded an R 2 of 0.76. Crop biomass was greatest in HD and HY with averaged yields of 7.8 at Yarrowitch and 2.2 tons ha−1 at Kirby. TI and PP had significantly lower biomass (average 4.6 and 0.7 tons ha−1 at Yarrowitch and Kirby respectively), especially PP at Kirby. Crop biomass production in CO was very small in both trials; 0.6 tons ha−1 at Yarrowitch and only 0.06 tons ha−1 at Kirby. Compared with HD, there was no significant crop loss in HY, but biomass was reduced by about 50% in TI and PP and 93% in CO.
Lettuce growth
The relationship between relative lettuce cover and time was fitted to a linear model (Fig. 7). The treatment effects varied significantly between trials, although HD and HY generally had the highest growth rates (i.e. slope coefficients), PP and CO had the lowest growth rates and TI was intermediate. In 2000 at Kirby, when weed pressure was the lowest, the lettuces in the CO showed equivalent growth to the other weed control treatments (P=0.10). Final relative crop cover for HD was greatest at Yarrowitch (~70%), and between 25 and 40% in the other trials. A test of the effect of cumulative rainfall on relative cover indicated that, like echinacea, lettuce growth was affected by rainfall received (P<0.001) and that supplementary irrigation, as used in commercial production, is needed to maintain lettuce yields. Despite the low weed levels in PP (Fig. 4), lettuce growth was reduced in both trials, especially at Kirby. Lettuce fresh weight biomass at harvest (raw data not shown, see Fig. 8A for derived values) had the same treatment rankings across the four trials as relative crop cover, and the two variables had a correlation of R 2=0.66. Lettuce yields for HD, TI and HY in these trials were similar to commercial yields, despite the lack of irrigation. The highest yields were observed in HD and HY, with lettuce biomass averaging ~23 tons ha−1. Yields in TI were ~13% less than HD and HY (P⩾0.089), while CO was reduced by 25% and PP by ~50%.
Effect of treatments on relative lettuce cover. The circles show the data points, the solid lines are the linear regression curves, the shaded area represents the 95% confidence limits and the equations describe the regression for each treatment. The PP treatment was not used in 2000.
Mean and standard error of GCV (A) and ACV (B) of echinacea for the weed control treatments. Note different scales on horizontal axes.
Weed–crop relationships
The relationship between weed variables (relative cover, density and biomass) and crop variables (relative cover and biomass) at echinacea harvest were assessed using linear regression (Table 2). Relative weed cover at Yarrowitch had the highest R 2 values with the crop variables and weed density was also high. Relative echinacea cover had higher R 2 values with the weed variable than echinacea biomass. The PP treatment had low weed levels but also had lower crop yields, so excluding the PP treatment improved the goodness of fit. At Kirby, the weed cover relationships were weaker, but excluding PP gave R 2 values of 0.67 or higher.
Correlation (R 2) between weed variables for echinacea at harvest.
1 Values in brackets are the correlation when the PP treatment was excluded from the calculation.
The relationship between weed variables and lettuce yield variables at harvest were low in all trials (Table 3). Some treatments with low relative weed cover had lower lettuce yields (e.g., PP), while other treatments with moderate to high weed levels, such as CO and TI, may have had reasonable yields. Underlying weed levels did not consistently influence the relationships, with lower weed loads at Laureldale and Kirby in 2000 being associated with high and low R 2 values respectively.
Correlation (R 2) between weed variables for lettuce at harvest.
1 Values in brackets are the correlation when the PP treatment was excluded from the calculation.
Economic analysis
The direct cost of carrying out each weed control treatment is shown in Table 4. The most expensive treatment for both echinacea and lettuce was PP, at a cost of about $12,500 ha−1 (averaged across all trials), of which 65% was due to purchase costs, 30% for laying and <5% for follow-up hand weeding. In echinacea, the next most expensive treatments were HD at $9600 ha−1 and HY at $8900 ha−1. The costs for HY were based on purchase costs (14%), laying the mulch (50%) and supplementary hand weeding (36%). In lettuce, the cost of HY remained high ($7600 ha−1), but the HD costs were considerably lower ($4400 ha−1). The TI treatment was relatively cheap ($4000 and $985 ha−1 for echinacea and lettuce respectively), made up of 5% for machinery costs and 95% for labor (tractor driving and supplementary hand weeding). No direct costs were incurred in the control.
The cost of weed control treatments (AU$ ha−1) used in the echinacea and lettuce trials.
The GCV and ACV for echinacea are shown in Figure 8. In Figure 8A, the response pattern for GCV follows that reported for crop biomass, with large differences observed between trials and treatments (P<0.001). HD and HY produced similarly high GCV, averaged at $49,000 ha−1 and $15,500 ha−1 at Yarrowitch and Kirby respectively, whilst the GCV for TI was reduced by half and CO about 90%. PP performed moderately well at Yarrowitch, with a GCV about one third less than HD and HY, but performed poorly at Kirby, with GCV reduced by about 70%. For ACV (Fig. 8B), treatment rankings were unchanged compared with GCV at Yarrowitch, but not at Kirby. In the latter trial, the significantly lower crop growth reduced the yield advantages observed at Yarrowitch. The ACV for HD and HY were equivalent to the cheaper TI treatment but more cost effective than CO (P⩽0.052). A loss of about $9000 ha−1 was incurred for PP at Kirby due to the relatively high cost of the mulch and the poor crop yield.
The highest GCV for lettuce was recorded at Yarrowitch ($19,000 ha−1 averaged across treatments), followed by Kirby in 2000 ($17,000 ha−1) and the lowest GCVs were observed in the trials at Laureldale ($9400) and Kirby in 2001 ($14,200 ha−1) (Fig. 9A). Averaging the treatment responses across the trials, HD, HY and TI had the highest GCV, with values for CO slightly lower (P=0.014), and considerably lower for PP.
Mean and standard error of GCV (A) and ACV (B) of lettuce for the weed control treatments. Note different scales on horizontal axes.
The highest ACVs were recorded for TI, CO and HD, with an average for the three treatments in all trials of $13,200 ha−1 (Figure 9B). HY had a mean ACV of $8700 ha−1, significantly lower than TI and CO (P⩽0.015) but not HD (P=0.114). PP had very low ACVs, with a mean loss of $4100 ha−1 across trials. The ACVs showed a similar pattern amongst treatments at Yarrowitch and Kirby in 2001. The top four treatments in those trials, TI, HD, CO and HY, were similar. However different rankings were observed in the trials in 2000. At Laureldale, the ACVs for CO and TI were significantly higher than HD and HY, reflecting their lower costs. The response at Kirby in 2000 was unique amongst the trials with CO being the most cost effective treatment. TI and HD were equivalent to each other (P=0.493) and HY was slightly lower (P=0.032).
Discussion
The HD treatment was very effective for weed control in lettuce and echinacea and gave excellent crop yields. It is the most widely used method of weed control amongst organic herb and vegetable growers in the industrialized worldReference Walz1, Reference Kristiansen, Jessop, Sindel, Rowe, Mendham and Donaghy8 and amongst many farmers in the developing worldReference Chatizwa20, although it fails to draw much attention in several reviews of non-chemical weedingReference Bàrberi21, Reference Bond and Grundy22. Hand weeding is usually considered expensiveReference Melander, Foguelman and Lockeretz7, although not always, especially in areas with cheaper labor costsReference Alemán23. The amount of time used in each application of the HD treatment (198–228 and 268–312 h ha−1 for lettuce and echinacea respectively) was within the range of reported times for various vegetable crops, which include 100–300 h ha−1 in organic carrotsReference Rasmussen, Ascard, Glen, Greaves and Anderson24, 256 h ha−1 in echinaceaReference Buntain6 and up to 500 h ha−1 in organic onionsReference Melander, Foguelman and Lockeretz7. One thorough hand weeding in lettuce was sufficient to prevent most weed growth up to harvest. While lettuce has several uncompetitive traits such as low growth habit, small root system and high nutrient and water requirementsReference Gallardo, Jackson and Thompson25, the short growing season and the rapid early growth of the transplanted lettuce seedlings enable the crop to compete effectively against weeds and achieve good yields with a single weeding eventReference Roberts10, Reference Weaver26. Echinacea has a longer growing season, with limited growth in the first two months when the crop is very susceptible to weed competitionReference Buntain6. Three weeding operations at 4, 8 and 12 WAP in the echinacea trials were required to maintain an adequate level of weed control up to harvest. Once shoot and leaf growth accelerate, echinacea becomes more competitive against weeds and can develop a dense, dominant canopy. The management emphasis should be on maintaining a high level of weed control up to about 10 WAP in echinacea.
Hay mulch gave good weed control and crop yields, although laying the mulch was time-consuming and some follow-up hand weeding was necessary. Hay and straw mulches have been successfully used for weed control and have generally produced good crop yieldsReference Olsen and Gounder4, Reference Alemán23. However, variable or negative responses have also been reported, commonly due to the prevailing weed populations, varying decomposition rates, the natural variability of the mulch and unevenness of layingReference Munn27, Reference Stirzaker and Bunn28. These trials were generally free of aggressive rhizomatous weeds common in organic horticulture in Australia such as Cynodon dactylon L. (Pers.) and Rumex L. speciesReference Kristiansen29, although annual weeds were able to germinate and emerge. Light penetration through the hay treatment is unlikely to have been completely prevented. Teasdale and MohlerReference Teasdale and Mohler30 found that 8 and 16 tons ha−1 of ryegrass mulch (cf. 9.5 tons ha−1 for the HY treatment reported here) reduced incident light by 90 and 95% respectively, but that did not prevent germination in phytochrome-dependent weed seeds. Hay mulch was quite cost effective for echinacea, but was less economic for lettuce. The high labor input for laying mulch and supplementary hand weeding would only be justified in high-value crops and where the weed flora were suitable (e.g., rhizomatous weeds absent or few). In intensive organic production in Australia, heavy continuous mulching is often practiced. However, on larger farms where tractors are used to prepare the ground, permanent mulch cover is uncommonReference Kristiansen, Jessop, Sindel, Rowe, Mendham and Donaghy8.
To overcome difficulties with laying bulky organic mulches, several options exist for mechanizing the management of mulches. These alternatives include roll-on mulches using existing plastic mulch technologyReference Olsen and Gounder4, flowable solidsReference Smith, Gilliam, Edwards, Eakes and Williams31 and in situ cover crop mulchesReference Creamer and Dabney32. These methods are often expensive and require skilled management. PP was used in these trials as an alternative to hay mulches. While PP was more expensive to buy than hay, it was in a form that enabled quicker and more even application to the planting beds, and required less follow-up hand weeding. Several reports have noted better weed control by a newspaper mulch than mulches composed of crop residuesReference Munn27, Reference Monks, Monks, Basden, Selders, Poland and Rayburn33. A high level of weed control was achieved due to the thick, continuous layer that prevented light penetration for germination and weed emergence. In addition to surface area, the available space within a mulch is an important propertyReference Teasdale and Mohler34. The PP had very little space between particles compared with the hay mulch, especially after wetting.
Although the PP suppressed weeds effectively, crop yields were reduced, particularly lettuce. Lower yields under PP are commonly attributed to soil nitrogen immobilization in response to the high carbon:nitrogen ratio of the mulchReference Monks, Monks, Basden, Selders, Poland and Rayburn33, Reference Guertal and Edwards35. A separate investigation found no evidence of phytotoxicity and leaf tissue nitrogen for lettuce was significantly lowerReference Kristiansen29. The use of such a mulch is very limited unless the nutrient balance (C:N ratio) is addressed, for example by adding manure. The high purchase cost is also prohibitive in the current commercial environment. However, technical and policy developments such as improvements in feedstock (e.g., paper waste) availability and processing, and changes in recycling requirements by government authorities, may improve this situationReference Olsen and Gounder4.
Apart from the CO treatment, tillage was the least effective method of controlling weeds. Satisfactory weed control was usually achieved in the interrows but not in the crop rows where the competitive effects of weeds would be greatest. The treatment was more effective in sandier soils (e.g., Yarrowitch) where better weed displacement was achieved and in light weed loads (e.g., Kirby in 2000). The dominant weeds at all sites were mostly annuals that benefited from the transition from pasture to annual cropping due to their large numbers, precocity and fecundity, except A. vulgaris at Kirby. But A. vulgaris benefited from cultivation by the extensive redistribution of its root system and shoots along the planting beds. Individual tillage operations were not expensive, but the on-going costs accumulate over time, and extra hand weeding may be necessary. In a longer-season crop such as echinacea, repeated tillage operations were needed, reducing cost effectiveness compared with the other treatments. However, in a short-season crop like lettuce, a single tillage operation was sufficient, so tillage costs were very low. The implements used in the trials were basic chisel ploughs with limited ability to accurately control weeds close to the crop plants without damaging crop roots. While this type of implement is commonly used by organic herb and vegetable growers in AustraliaReference Kristiansen, Jessop, Sindel, Rowe, Mendham and Donaghy8, more accurate, though more costly, implements are availableReference Welsh, Tillett, Home and King36. Other research has also found that tillage can be less costly than hand weedingReference Melander, Foguelman and Lockeretz7, Reference Alemán37 and mulchingReference Edwards, Shuster, Huelsman and Yardim38, Reference Litterick, Redpath, Seel, Leifert and Marshal39.
The unweeded control treatment provided a measure of the pre-existing weed levels in each trial and a basis for evaluating the impact of the other treatments. In general, weeds were abundant in the control plots and crop yields were reduced, although lettuce showed potential as a quick cash crop without weeding when weed numbers were low. A study of radish (Raphanus sativus L.), a crop with a very short growing season, demonstrated that rapid crop development can avoid the need for weeding in short-season cropsReference Turner, Lennartsson, Bond, Grundy, Whitehouse and Marshal40, particularly if the starting weed load is light. Relying on minimal control techniques that are effective in the short term may compromise the longer-term goal of weed seed bank depletion. The consequences of weed seed production by uncontrolled weeds on management in subsequent seasons would also need to be considered by growers, with a number of options available including grazing, mowing or tilling.
Differences in weed levels between trials were due to variations in soil conditions and bed preparation, weather and previous land use. The heavy clay soil at Laureldale caused the planting beds to have larger clods than in the finer structured soils at Yarrowitch and Kirby. Large soil clods can reduce weed seedling emergence and crop establishment. As expected, trials with greatest weed density and final weed biomass also had the highest rainfall. The six-fold increase in weed density observed at Kirby between 2000 and 2001 (previously uncultivated) parallels changes in weed levels at Yarrowitch after converting from grazing to annual cropping (P. Green, personal communication) and reported elsewhereReference Sjursen41, Reference Belde, Mattheis, Sprenger and Albrecht42.
There was no correlation between the cost of the treatments and the crop yield in these trials. A survey on weed management amongst Australian organic herb and vegetable growers found that the perceived expense of a weed control method was not related to the perceived success of that methodReference Kristiansen29. Bond and LennartssonReference Bond, Lennartsson and Marshal43 note that relatively expensive methods may be cost effective in certain cropping situations, and that success depends on matching the weed control technique with the cropping strategies. Several other factors influence the economics of weed management (and therefore the choice of methods) including the size of cropping area, prevailing weed density, the cost of labor and commodity price fluctuationsReference Melander, Foguelman and Lockeretz7, Reference Alemán23. The economic analysis reported here was specific to the time and location of the trials, and outcomes will vary depending on the availability of labor, machinery and implements, and suitable mulch materials.
Conclusions
Cheaper weed control methods such as tillage with limited follow-up hand weeding may be sufficient to ensure a reasonable yield for lettuce, a crop with a short growing season and rapid early growth from transplanted seedlings. The acceptable economic return of the CO treatment suggests that good weed control in the cropping area prior to planting may even be adequate. More expensive weeding methods such as hand weeding and hay mulch are cost effective weed management options for longer-season and higher value crops like echinacea. Crop selection should be based on prevailing weed loads, as they will strongly influence the efficacy of the weeding methods chosen.
These trials indicate that hand weeding, tillage (even with unsophisticated implements) and crop residue mulches are cost effective in the appropriate setting. Integrating these methods (e.g. tillage and chipping), in association with other techniques such as strategic crop rotations, false seedbed preparation and fertilizer banding, may improve their effectiveness further.
Acknowledgements
Technical assistance was provided by Peter Green (Horticultural Consultant, Yarrowitch) and David Edmonds, Kevin Saunders and Faye Miller (Technical Officers, UNE). Ian Davies (Biometrician, UNE) and Duncan Mackay (Research Assistant, UNE) provided statistical advice. Funding from the Rural Industries Research and Development Corporation and UNE is gratefully acknowledged.
