Introduction
California's upland cotton production ranks third in the nation, generating about 10% of the United States' total1. Over the period 1996–2001, there were between 300,000 and 400,000 hectares of planted upland cotton in the state1. Over the period of 1999–2001, there were between 365 and 486 hectares of certified organic upland cotton, comprising less than 1% of the state's total2. Cotton is historically one of the top contributors to total farm income in California. From 1996 to 2001, the total value of the California cotton crop ranged from $918 million to $356 million, respectively3.
Cotton is also one of the largest users of agricultural chemicals of any commodity produced in California. In 2001, cotton growers applied 5.5% of all pesticides reported used in the state4. During this time, an average of over 10 kg of pesticide active ingredient (AI) was applied per hectare in California cotton4.
From 1994 through 1995, we compared organic and conventional cotton production in the Northern San Joaquin Valley (NSJV)Reference Swezey and Goldman5,Reference Swezey, Goldman, Jergens and Vargas6. During that project, we found that cotton growers were very interested in the potential environmental and economic benefits of pesticide use reduction strategies and organic cotton production. A study was initiated that compared three production systems: reduced insecticide input/integrated pest management (IPM), certified organic and conventionally managed cotton. Four key questions were addressed: (1) Would the reduction and/or elimination of insecticides impact pest and beneficial insect abundance? (2) Would the cost structure of each management strategy vary, and if so, how? (3) Would harvest parameters, such as lint quality and yield, differ based on management strategy? (4) Would organically grown cotton be viable and competitive with other local management strategies?
Methods and Materials
Plot selection and management
In April of each year from 1996 to 2001, fields were selected and designated as one of three treatments: organic, IPM and conventional (Table 1). Organically managed fields were registered with the California Department of Food and Agriculture Organic Program, certified according to the California Organic Foods Act (COFA) of 1990, and inspected by a third party (Quality Assurance International, Inc.). Fields designated as organic or IPM were managed by growers in the BASIC (Biological Agriculture Systems in Cotton) program. This grower-initiated program encouraged enrolled growers to plant their cotton early (in April), and near alfalfa fields for pest and beneficial insect attractionReference Sevacharian and Stern7. Growers in this program received weekly monitoring reports from the authors and BASIC personnel for insecticide use decision-making. These growers also received periodic early season releases of insectary-reared green lacewing (Chrysoperla carnea) larvae, at an average of 12,000 larvae per hectare, for management of spider mites and aphids. Reductions of calendar-based insecticide applications were also encouraged in IPM fields. Conventional fields were managed outside of the BASIC program.
Grower information and fields monitored from 1996 to 2001.
Fields were chosen based on geographic location (within an 8 km radius of Chowchilla, California), size (8–32 hectares) and planting date (mid-April to early May). Cotton plant variety was also used to select fields; San Joaquin Acala variety Maxxa was used exclusively in organic fields, and was used predominately in IPM and conventionally managed replicates from 1996 to 1999. Beginning in 2000, the genetically modified (Roundup Ready™) Riata variety of the Maxxa breeding line was used in nearly all IPM and conventional fields. Differences in specific production practices among each of the treatments are shown in Table 2.
Summary of production practices and inputs 1996–2001.
Arthropod pests and natural enemies
Both-sweep net samples and visual leaf counts were made for determination of arthropod abundance. Sampling was conducted weekly from June to September, during which time three distinct intervals of the production season were established: early (June), mid (July) and late (August–September). Fields were divided into four quadrants and samples of 50 sweeps were taken from each quadrant with a 38 cm canvas insect net. Each sample was taken from one randomly chosen row, and field edges (the outer ∼15 m of each field) were avoided. This procedure was used to collect the western tarnished plant bug (WTPB), also known as the lygus bug (Lygus hesperus Knight), which is a key cotton fruit pest, and generalist predators Orius tristicolor (minute pirate bug), Geocoris spp. (big-eyed bug), Nabis sp. (damsel bug), Zelus sp. (assassin bug) and Chrysoperla spp. (green lacewing). Insect abundances were recorded in the field using a small tape recorder. These records were later transcribed in the laboratory for entry into a database.
Visual leaf counts were conducted by weekly examination of the eighth main stem leaf from the plant apex. Five plants were chosen randomly from each of the four quadrants to give a total of 20 leaves sampled per field. Spider mites (Tetranychus spp.) and cotton aphids (Aphis gossypii) were recorded using presence/absence counts. Generalist predators such as minute pirate bugs, big-eyed bugs (eggs and nymphs), green lacewings (eggs and larvae) and western predatory mite (Metaseiulus occidentalis) were also counted in these leaf samples.
Plant development parameters
Plant development parameters were measured weekly from June to September. Five plants were chosen randomly and removed from each of the four quadrants, for a total of 20 plants per field on each sampling occasion. The architecture of these plants was mapped using guidelines from the University of California8, which were developed to help predict yields, measure plant stress and insect damage, and time growth regulator and defoliant applications. Parameters measured included retention of first position bolls (the highest yielding position) and the number of bolls in the bottom five fruiting positions (a measure of loss due to insect damage or other stress). Total bolls per plant were also measured immediately prior to harvest.
Plant density
Shortly after cotton plant emergence in each sample year, we determined the average plant density (plants per hectare) in each field. From each of the four quadrants, we counted and averaged the number of plants in four randomly selected 1/2470 ha linear row samples. This average was then multiplied by 2470 to determine the number of plants per hectare. Plant density was again determined at harvest each year.
Pesticides
The type, quantity and frequency of pesticide applications were recorded for each field treatment through grower interviews conducted after harvest each season, from 1997 to 2001. Pesticide quantities were converted into kilograms of AI per hectare based on manufacturer's label rates. Pesticides were classified as insecticides, miticides, herbicides, crop preparation growth regulators and defoliants.
Cost of production
Production costs were also obtained through post-harvest grower interviews in December to January of each following production year, excluding 1998. An average budget was then made for each treatment following the methods established by Klonsky et al. Reference Klonsky, Tourte, Swezey and Chaney9. Costs were divided into pre-harvest- and harvest-related categories. Included within each category grouping were labor (tractor operator hours, hand weeding), field power (ripping, leveling, disking, bedding, ditching and cultivating), materials (compost, water, seed, pesticides and lacewings) and custom/rentals (i.e., custom applications of compost or aerial spraying and harvest machinery rentals). Both total costs per hectare and cost per bale were calculated. When not directly provided by the grower, costs of individual operations were based on values presented by Klonsky et al. Reference Klonsky, Tourte, Swezey and Chaney9.
Cotton lint quality
Gin records were used to measure lint quality, an important determinant of lint price. Lint quality consists of several parameters, the most important of which are fiber length, fiber strength, fiber micronaire (diameter), color grade (a measure of lint staining caused by foreign material in the harvest) and bale leaf ratings (leaf contamination in harvested lint). We combined color grades into three groups, for ease of analysis. Grades 11, 21 and 31 generally receive the highest price, and were combined into the highest quality group. Groups two (grades 41, 51 and 61) and three (other) were lower quality lint. We obtained all gin records except for those from the organic fields in 1999, which were lost during a gin consolidation.
Cotton lint yield
Average lint yield (bales per hectare) per treatment was determined by using gin records obtained either from gin operators or during post-harvest interviews with participating growers.
Statistical analysis
Data taken throughout the production season, which included arthropod pests and natural enemies, plant development parameters and plant density, were analyzed using repeated measures analysis of variance (multivariate analysis of variance, MANOVA). Values were averaged by 2-week periods to address missing data. Values for missing data points were estimated by running a full analysis of variance (ANOVA) model with the data; the predicted values were then used in place of the missing ones. Tukey's Studentized Range (honestly significant difference, HSD) Test was then used for all post hoc comparisons, including growers' overall pesticide use, annual production costs per hectare and costs per bale, and yields. Lint quality differences were determined by single factor ANOVA (F ratio) tests. Analyses for plant development parameters, planting densities, lint quality (except for color grades) and yields were done using SAS Version 8.02 for Windows10. Analyses for insect abundance, pesticide use, costs of production and lint color grades were done using SPSS Version 11.511.
Results and Discussion
Arthropod pests and natural enemies
The only significant difference between treatments for total WTPB abundance (adults plus nymphs) was during the mid-season interval of 1998, when there were more total WTPB in organic than in conventional fields (P<0.05). Six-year average WTPB abundance was relatively low with reference to action thresholds, with mid-season averages of 2.83±0.85, 2.61±0.70 and 2.41±0.60 WTPB per sweep net sample, in organic, IPM and conventional treatments, respectively. These values were well below the suggested mid-season action threshold of 7–10 WTPB per sweep net sample12.
Significant differences in WTPB nymphal abundance between treatments were more frequent. Organic fields had more WTPB nymphs than did either IPM or conventional fields in the late season interval of 1996 and more than IPM fields in the early season interval of 2000 (P<0.05). Organic fields had significantly fewer WTPB nymphs than did IPM or conventional fields in the late season interval of 2000, and also fewer than conventional fields in the mid-season interval of 2001 (P<0.05). Again, 6-year WTPB abundance was relatively low, with mid-season nymph averages of 1.00±0.96, 0.86±0.81 and 0.64±0.29 bugs per sweep net sample, for organic, IPM and conventional treatments, respectively.
There were no consistent differences in WTPB abundance, between any treatments. During this study, WTPB did not reach pest status in any treatment in the NSJV. This result could be due to the admixture of alfalfa fields (a preferred host) in this area of cotton productionReference Sevacharian and Stern7,Reference Stern, van den Bosch and Leigh13–Reference Godfrey and Leigh15.
Mean percentage mite infestation was significantly greater in organic fields than in conventional fields during the early season intervals of 1996 and 2000, and greater than IPM fields in the early season interval of 1996 (Fig. 1). In 1996 and 1999, mite infestation in organic fields slightly exceeded 50%, the upper end of the action threshold12. Such infestations can impact subsequent yields8, and occurred in spite of sulfur use by organic growers (see pesticides below). A typical whole field application of sulfur for mite control was made at the label rate of 34 kg ha−1. However, in this study, organic growers more frequently applied 0.9–1.4 kg of dusting sulfur on field margins only. Sulfur use in organic fields (11.8 kg ha−1) constituted the largest pesticide input recorded in any treatment. Beginning in 2000, organic and IPM growers lowered mite infestation levels by better utilizing monitoring information, which enabled them to apply timelier synthetic miticide or sulfur treatments.
Average percentage mite infestation from ‘early’ season (June). Percentages based on mite presence/absence counts on cotton plants. Different letters within years represent statistical differences between treatments (ANOVA; P<0.05).
Percentage cotton aphid infestation was significantly greater in conventional fields than in organic fields during the late season interval of 1996, and greater than both organic and IPM fields during the late season interval of 1999 (Fig. 2). The lack of early or mid-season insecticide stress in organic cotton may have favored naturally occurring biological control of cotton aphidReference Wilson, Bauer and Lally16. High levels of late season cotton aphid infestation can cause aphid honeydew accumulation on open bolls, resulting in sticky cotton lint. This is problematic during harvest and ginning, and can also lead to fungal growth, which can stain the cotton fibers. No sticky cotton was reported in gin records in any year of the study, which suggests aphid pressure remained below economic thresholds.
Average percentage aphid infestation from late season (August–September). Percentages based on aphid presence/absence counts on cotton plants. Different letters within years represent statistical differences between treatments (ANOVA; P<0.05).
Total generalist predator abundance (adults plus nymphs) was significantly greater in organic fields than in either conventional or IPM fields, during ten and four different seasonal intervals, respectively (Table 3). IPM fields also had greater total predator abundances than did conventional fields during the mid-seasonal intervals of 1997 and 1998.
Average total natural enemy abundance (±SEM) by season (1996–2001).
Different letters within seasonal intervals represent statistical differences between treatments (ANOVA; P<0.05).
Nymphal natural enemies were significantly more numerous in organic fields than in conventional or IPM fields during seven and three different seasonal intervals, respectively (Table 4). IPM fields also had greater abundances of predator nymphs than did conventional fields during the mid- and late seasonal intervals of 1996.
Average nymphal natural enemy abundance (±SEM) by season (1996–2001).
Different letters within seasonal intervals represent statistical differences between treatments (ANOVA; P<0.05).
Generalist predator density on cotton leaves was significantly greater in organic fields compared with conventional fields in the early and mid-season intervals of 1996, and the mid-season interval of 1999 (P<0.05). These predators in organic fields were also more abundant than those in IPM fields during the mid-seasons of 1996, 1999 and 2001, and the late season of 2001 (P<0.05). Six-year average leaf-sampled predator densities were 0.22±0.04, 0.16±0.02 and 0.12±0.02 per leaf, for organic, IPM and conventional treatments, respectively.
Organic fields had significantly more total natural enemies, nymphal natural enemies and leaf-sampled predators than did conventional fields in 56% (10/18 sampled months), 39% (7/18 sampled months) and 17% (3/18 sampled months) of recorded seasonal intervals. This pattern of higher abundance of natural enemies in organic than in conventional cotton fields is confirmed in other organic crops grown in California, including strawberriesReference Gliessman, Swezey, Allison, Cochran, Farrell, Kluson, May and Werner17 and tomatoesReference Letourneau and Goldstein18. There was no recorded seasonal interval in which conventional fields had a significantly greater abundance of natural enemies than either organic or IPM fields.
Plant development parameters
There were no significant treatment differences between any in-season plant development parameters (P>0.05). The 6-year average number of first position bolls (the highest yielding position) was 7.42±0.53, 7.39±0.75 and 8.18±0.53 for organic, IPM and conventional treatments, respectively. The percentage boll retention at the five lowest positions (a measure of possible WTPB damage to early fruit) was 65.40±6.29, 68.00±3.58, and 63.80±3.17 for organic, IPM and conventional treatments, respectively. The total number of bolls produced per plant at the end of each year was 7.93±0.41, 6.78±0.55 and 7.29±0.46 for organic, IPM and conventional treatments, respectively. The lack of differences between treatments with respect to percentage retention of bolls at the bottom five positions and total bolls at harvest is reflective of the consistently low WTPB pressure that we found in all treatments.
Plant density
Organic fields had significantly fewer plants at harvest than conventionally managed fields from 1996 to 1999 (Fig. 3). Organic growers generally planted their fields at lower densities during these years, believing that lower plant densities would result in less leaf contamination and generate greater boll formation. Density differences at harvest were therefore not surprising. Organic plant densities at harvest were also significantly lower than IPM fields from 1997 to 1998. Beginning in 2000, differences in plant density were not significant between treatments, as conventional and IPM growers reduced densities to levels comparable with organic fields.
Average cotton plant density per hectare at harvest (1996–2001). Different letters within years represent statistical differences between treatments (ANOVA; P<0.05).
Pesticides
Over 6 years of production, organic fields were treated with none of the synthetic pest control inputs typical of conventional and IPM fields, such as synthetic insecticides and miticides, herbicides, growth regulators and defoliants (Table 5). However, the AI weight of ‘total pest control materials’ and ‘total pesticides’ did not differ between organic and conventional treatments. This was due to the use of wettable and dusting sulfur (a miticide) in organic fields, which was the only pesticide used by organic growers. The volume of sulfur use was also responsible for organic fields accumulating higher weight of AI of ‘total pest control materials’ than IPM fields.
Average pesticide use (1997–2001) in kilograms of AI per hectare (±SEM).
Different letters within categories represent statistical differences between treatments (ANOVA; P<0.05).
Synthetic insecticide use in IPM fields was significantly lower than in conventional fields.
Insecticide use in IPM fields averaged 0.63 kg AI ha−1, as opposed to 1.02 kg AI ha−1 for conventional fields, a reduction of 38%. Use in all other pesticide categories was not significantly different between these two treatments.
Cost of production
Six-year average operational costs of production per bale were significantly higher for organic than for IPM or conventional fields (Table 6). These costs were 37% higher for organic than for conventional fields. The increased costs per bale were mainly due to lower yields in the organic system (see below). When calculated per hectare, costs of production were not statistically different, being only 5 and 3% higher for organic and IPM fields, when compared with conventional fields (P>0.05). Operational differences included higher costs for labor (hand weeding crews and cultivation), custom applications (compost and sulfur) and harvest field power (increased harvest time and effort) in organic fields and higher materials costs (synthetic insecticides) in conventional fields. Costs of production per bale for IPM fields were more variable and not significantly different than conventional costs.
Average operational production costs (±SEM) from 1996 to 2001 ($).
Different letters within categories represent statistical differences between treatments (ANOVA; P<0.05).
Cotton lint quality
Cotton fiber length, strength and micronaire (fiber width) did not differ between treatments in any year (P>0.05). Overall fiber lengths (1/81 cm) were 36.13±0.18, 35.68±0.25 and 35.99±0.22 for organic, IPM and conventional treatments, respectively. Fiber strength values (gtex) were 32.97±0.41, 31.94±0.25 and 32.03±0.41 for organic, IPM and conventional treatments, respectively. Micronaire or fiber width values (μm) were 4.24±0.05, 4.25±0.08 and 4.08±0.05 for organic, IPM and conventional treatments, respectively.
The percentage of bales that were of the best color grades of lint (11, 21 and 31) was significantly lower in the organic than in the conventional treatment in 1997, 1998 and 2001 (Fig. 4). This combined color grade series was also significantly lower in the IPM than in the conventional treatment in 1997 and 2001.
Average percentage cotton lint color grades (1996–2001). The grades 11, 21 and 31 represented the highest quality fiber, while 41, 51 and 61 represented the lowest. Different letters within years represent statistical differences for the grades 11, 21 and 31 between treatments (ANOVA; P<0.05). C, conventional; I, IPM; O, organic.
Bale leaf ratings from organic fields were significantly higher (reflecting more leaf contamination) than ratings from conventional fields in 1996 and 1997, and significantly higher than ratings from IPM fields from 1996 to 1998 (Fig. 5). Bale leaf ratings were significantly higher in conventional relative to IPM cotton in 1996 and 1998.
Average bale leaf ratings (1996–2001). Higher ratings represent greater cotton leaf contamination in the harvested lint. Different letters within years represent statistical differences between treatments (ANOVA; P<0.05).
These differences in color grade and bale leaf rating were likely due to insufficient pre-harvest defoliation in organic fields, rather than differences in cotton boll development, as illustrated by non-significant treatment differences for cotton lint length, strength and width. The mineral and organic acids used as late-season nutritional crop preparation aids by organic growers were much less effective than the common chemical defoliants used by IPM and conventional growers in preventing plant material from contaminating lint. This lack of effective defoliation also required slower harvest machinery operation, and occasional repeated mechanical pickings.
Cotton lint yield
Cotton lint yields were significantly higher in conventional than organic fields from 1997 to 2001 (Fig. 6). A small sample size decreased our ability to detect treatment differences in 1996. Yields from conventional fields were also significantly higher than those from IPM fields in 2000. Average 6-year yields were 4.4, 5.4 and 6.7 bales ha−1 for organic, IPM and conventional treatments, respectively. Significant 6-year average yield differences between organic and conventional fields were 1.3 bales ha−1 (19% reduction) in IPM fields and 2.3 bales ha−1 (34% reduction) in organic fields, relative to yields in conventional fields (P<0.05). Treatment yields in IPM and conventional fields consistently tracked average cotton yield trends for Madera and Merced counties during this study period.
Average cotton yields from 1996 to 2001. One bale=218 kg of cotton. Different letters within years represent statistical differences between treatments (ANOVA; P<0.05). County values are averages of Madera and Merced counties, California.
Less than optimum weed management in organic fields may best explain the yield differences between conventional and organic treatments. Weed densities in organic fields were observed to be higher than either IPM or conventional fields in several years of this study. For example, johnsongrass (Sorghum halepense) is one of the most commonly encountered weeds in the NSJV, and is capable of reducing cotton yields by up to 90%Reference Keeley and Thullen19. Competition from other weeds in the NSJV, such as cocklebur (Xanthium spp.), amaranth (Amaranthus spp.) and pigweed (Amaranthus spp.) have also been shown to reduce cotton yieldsReference Snipes, Buchanan, Street and McGuire20–Reference Showler and Greenberg22.
Methods for weed control available to organic growers are limited and are the most likely cause of higher weed densities. While machine cultivation is available and utilized by all growers during the earlier stages of cotton vegetative growth, only IPM and conventional growers use early season herbicide applications in tandem with herbicide-resistant cotton (beginning in 2000) to effectively reduce weed densities. Early season machine cultivation alone is not sufficient to suppress weed populations later in the growing season. Given the size and structure of cotton plants during later stages of development, machine cultivators are unable to remove weeds without damaging the crop. Therefore, organic growers become reliant on hand weeding, which can be cost-prohibitive.
Given the lack of significant differences between treatments for such cotton production parameters as the number of first position bolls and total bolls, increased weed pressure may best explain yield differences. Strategies that may raise yields in organic fields include timelier hand weeding, more frequent early season cultivation and adoption of innovative weed control technologies (flamers, organically compliant contact herbicides, etc.).
Summary and Conclusions
There proved to be two major future obstacles to the viability of growing cotton organically in the NSJV of California. Higher production costs, coupled with lower yields, established an economic necessity for a price premium for organic cotton. However, an abundance of relatively inexpensive overseas cotton (both organic and conventional) contributed to diminished market prices available to California cotton growers from 1996 to 2001. During this study, global cotton market prices reached a 30-year record low ($0.97 kg−1)1. Organic growers participating in this study required as much as $2.42 kg−1 to ensure a profit after accounting for storage, brokerage and other opportunity costs (Claude Sheppard, personal communication). From 1999 to 2004, there was a 78% reduction in California certified organic cotton hectares2. In 2004, approximately 41 ha of registered organic cotton remained in the NSJV (in Madera County), and only 111 ha remained statewide2. The discrepancy between costs and prices awarded illustrates the current lack of marketplace-mediated incentives to encourage certified organic production practices. There are also no current means to integrate environmental costs into the market price of conventionally managed cotton, which would also help to bridge this price gap.
Synthetic insecticide applications were reduced by IPM growers, who applied 38% fewer kg AI per hectare than did conventional growers. Organic growers used no synthetic insecticides. However, in the majority of comparison intervals, pest insect abundance in organic or IPM fields was equivalent to abundance in conventional fields, and only rarely exceeded action thresholds. This reduction and/or elimination of insecticides also conserved beneficial insects in the IPM and organic fields, particularly in the latter. This study demonstrated that both a supervised IPM program and certified organic production resulted in successful insect pest management in cotton from 1996 through 2001, while substantially reducing insecticide applications in the NSJV.
There were significant differences in the cost of production between management strategies. Six-year averages of costs per bale were significantly higher for the organic than for the IPM or conventional treatments. This increase in costs in the organic treatment (37% greater than conventional costs) was mainly due to lower yields, higher costs for labor (for hand-weeding crews) and pre-harvest custom applications (of compost and sulfur). Higher materials costs (synthetic insecticides) were accrued in conventional and IPM fields.
Cotton yields were also significantly lower in organic fields than either IPM or conventional fields. The lower yields in organic fields were likely due to lower plant densities and higher weed pressure. Yields from IPM fields were significantly lower than those from conventional fields, although the difference was smaller.
This multi-year study illustrates the potential for insecticide use savings in IPM and organic cotton, and the yield potential for these alternative production systems in the NSJV. Organic cotton in the NSJV had lower yields and higher costs of production. Low world market prices that prevent premiums from being offered for NSJV organic cotton are the primary obstacle to its continued production.
Acknowledgements
We would like to thank Claude and Linda Sheppard, Pete Cornaggia and other cotton growers for participating in this study, and the BASIC program for helping to recruit growers for participation. We would also like to thank Richard Nathanson for assistance with data analysis, and John Bailey, Merrilee Buchanan, Jennifer Charles, Amy Griggs, Amanda Lewis, Eri Mizuno, and BASIC program personnel for their assistance with fieldwork. This research was funded in part by the US-Environmental Protection Agency (EPA) Region IX grant no. NP-98932201-1. We thank Paul Feder and James Liebman (US-EPA Region 9) for their assistance with this project.
