WEED MANAGEMENT IN LOW-EXTERNAL-INPUT AND ORGANIC FARMING SYSTEMS

MATT LIEBMAN, LAMMERT BASTIAANS1, and DANIEL T. BAUMANN2 1

Iowa State University, Department of Agronomy, Ames, IA, 50011-1010, USA; Wageningen University, Department of Plant Sciences, Crop and Weed Ecology Group, P.O. Box 430, 6700 AK, Wageningen, The Netherlands; 2Swiss Federal Research Station for Fruit-Growing, Viticulture, and Horticulture, CH-8820, Wädenswil, Switzerland

During the past quarter century, growing numbers of government policy makers, scientists, consumers, and farmers have expressed concerns over the impacts of conventional farming practices on environmental quality, human health, and the economic viability of farm families and rural communities. Particularly in western Europe, and to a certain extent in the USA, Canada, and other countries, these concerns have begun to translate into changes in public policy, research priorities, and market opportunities that favor the development of low-external-input (LEI) and organic farming systems. For example, during the 1980s in Sweden, a >50% reduction in agricultural pesticide use was mandated and achieved through coordinated sets of regulations, research and extension education activities, and economic incentives (Bellinder et al., 1994; Matteson, 1995). Similar approaches have been initiated in other European countries (Matteson, 1995). Concurrently, consumer demand for organic crop and livestock products has grown 20 to 25% per annum in the USA, many European countries, and Japan (Geier, 1998; Myers and Rorie, 2001). Sales of organic products in 2000 were estimated at $7.8 billion in the USA (Myers and Rorie, 2001) and at least $7.5 billion in Europe and the UK (Soil Association, 2001). Continued growth is expected in the amount of land managed with LEI and organic methods and in associated economic activity. Weed management is a focal point for all farming systems, but it is especially critical in LEI and organic systems. The avoidance or exclusion of herbicides in these systems can lead to situations where weed density and weed competition against crops rise to unacceptable levels unless large amounts of mechanical cultivation or hand labor are used (Vereijken, 1994; Rasmussen and Ascard, 1995). Heavy reliance on both of these options is undesirable, however, since frequent cultivations can reduce soil quality and crop health, and hand labor for weeding generally has high cash and opportunity costs. If requirements for cultivation and hand weeding are to be minimized in LEI and organic farming systems, a better understanding of weed ecology and the full range of factors regulating weed density, growth, and competitive ability is needed (Liebman and Gallandt, 1997).

Inderjit (editor) Weed Biology and Management, pp. 285-315. © 2003 Kluwer Academic Publishers, The Netherlands.

286

Liebman, Bastiaans and Baumann

This chapter examines existing and emerging approaches for weed management in LEI and organic farming systems. After first discussing information requirements for weed management in such systems, tactics that can serve as components of multifaceted weed management strategies are described. Some use of herbicides may be included in LEI systems, but the emphasis in this chapter is on tillage options, soil fertility and moisture regimes, crop residue management practices, methods to improve crop competitive ability, cropping system diversification, and other tactics whose refinement is likely to make reductions in herbicide use more viable in LEI systems and to improve the performance of organic systems. Three case studies from Switzerland, the USA, and the Netherlands are presented that illustrate how combinations of tactics may be put to work against weeds in different LEI and organic farming systems. These case studies include a look at some of the shortcomings and remaining challenges of weed management. The chapter ends by identifying high-priority topics for future research in LEI and organic systems. The data and examples used in this chapter are drawn almost exclusively from temperate areas, with which we are most familiar, but we believe there are ample opportunities to develop LEI and organic systems in tropical regions as well. INFORMATION REQUIREMENTS Conventional weed management systems based on herbicides are designed for use in many different locations against broad spectra of weed species. In contrast, weed management systems that avoid or exclude the use of herbicides are more site-specific and require more careful consideration of the impacts of management tactics on individual weed species and weed community composition. Although general principles can be identified, actual weed management practices must be chosen, integrated, and adapted locally based on location-specific information. Information requirements for the design of weed management strategies in LEI and organic farming systems occur at several organizational levels. First, information must be assembled concerning key components and functions of the agroecosystem that affect and are affected by weed management. Next, the abundance, distribution, and ecological characteristics of key weed species present within the agroecosystem must be determined. Analysis of this information should result in the identification of weed management priorities, after which different tactical options can be selected and tested with respect to their efficacy and suitability within the specific agroecosystem. After evaluation these strategies can be adjusted and refined.

Weed Management in Low-External-Input and Organic Farming Systems

287

Agroecosystem Characteristics A large number of materials, organisms, and processes collectively define and determine the character of an agroecosystem. Within this complexity are certain key environmental factors, pieces of infrastructure, species, activities, and relationships that shape the context for developing and implementing weed management strategies. By studying the system-level context for weed management, key constraints and best opportunities for desirable changes can be identified by farmers and the researchers and extension professionals working with them. Conway (1985) and Conway and McCraken (1990) have described techniques for agroecosystem analysis in detail. The following factors are among those that could be included in agroecosystem analyses that have an explicit focus on weed management. Soil Characteristics Soil characteristics of different fields within a farm affect the timing of tillage and cultivation operations that kill, damage, or move weed plants and propagules. Soil characteristics also affect the availability of nutrients and water to crops and weeds, and the choice of crops within rotation sequences. Seasonal climatic patterns also determine when tillage and cultivation operations are possible, as well as the timing of crop and weed growth and resource use. As discussed later, the sequence of crops grown on different fields and the associated crop management practices strongly influence opportunities for weed recruitment, growth, and reproduction. The types of crops that a farmer can consider producing are determined, in many cases, by the opportunities that exist for marketing them and by government programs, such as subsidy payments. LEI and organic products may be directed into special marketing channels and may receive price premiums in the marketplace or special government subsidies. Integration of crops with livestock production determines not only the types of crops grown on the farm and within rotation sequences, but also manure management practices that affect weed seed survival and dispersal within and among farms. The equipment available for weed control and crop production affects the range of options for suppressing weeds and establishing and maintaining crop species. Equipment may be owned by the farm operator, or rented or borrowed within time constraints imposed by commercial firms and other farmers. Labor used for weed control and other farm operations may be provided by the farm family itself, by members of the surrounding community, or by contracted workers brought in from longer distances. Typically there are labor bottlenecks within the year during which labor requirements for weed management may conflict with labor needs for other onfarm and off-farm activities. Weed control options are also affected by the farmer’s financial situation, access to technical information, interactions with people with experience, and short-term and long-term goals. Often agroecosystem characteristics appear to be unchangeable and trivially obvious. However, closer analysis may reveal opportunities to implement desirable changes that were not immediately apparent. For example, analysis of yearly crop production patterns and climatic conditions may indicate when weed-suppressive cover crops could be introduced into rotation sequences. Successful management of a cover crop and its residue for weed suppression will depend on the availability of appropriate equipment, access to relevant information, impacts on additional pests,

288

Liebman, Bastiaans and Baumann

and other factors that may be identified by dialogue among farmers, scientists, extension personnel, agricultural consultants, and others. Weed Abundance, Community Composition, and Ecological Characteristics To develop and improve a weed management strategy for a given farm or set of farms, knowledge of key components and processes of the target agroeocosystem must be complemented with information concerning weed abundance and community composition. Because the individual weed management tactics commonly used in LEI and organic farming systems typically are less effective and less persistent than herbicides, there is a particular need in these systems to match specific weed problems with sets of tactics applied at specific times. Sampling to determine weed abundance and community composition serves additional purposes, which are not restricted to LEI and organic systems. Monitoring these parameters can indicate whether weed suppression tactics are working over short- and long-term horizons, whether changes in control tactics need to be initiated at particular locations on a farm or at particular times within a crop sequence, and whether “new,” noxious species have arrived and need to be controlled before they become widespread. Accurate assessments of weed density also are needed for threshold-based management strategies, but the details and effectiveness of such strategies remain the subjects of considerable debate. Methods to track weed abundance and community composition in farmers’ fields range from hand-drawn maps produced by visually assessing the vegetation at one’s boot-tip at several stops during a walk across a field, to computer-drawn maps produced by counting weeds within quadrats at numerous locations that are recognized by satellite-assisted global positioning systems. The choice of methods in a given situation will depend on the level of precision needed and the time, money, labor, and equipment that are available. Regardless of the actual method used, however, the dominant weed species within fields should be noted, the general patterns of their spatial distributions should be identified, and some repeatable estimate of their abundance (e.g., leaf area coverage of the soil, stem density, etc.) should be made. After the major species of the weed community have been identified, key aspects of their ecology that affect their responses to management tactics need to be determined through discussions, library research, and experiments. Identification of temporal patterns of emergence, growth, and reproduction of dominant weed species is useful for assessing the potential impacts of control measures imposed at different times. Consideration of weed phenological patterns can indicate which species are likely to be suppressed by particular control methods and which species are likely to escape unless other control measures are used. Similarly, knowledge of speciesspecific responses to burial at different depths in the soil can indicate how long a given weed is likely to persist as a seed or vegetative structure on or in the soil, and whether it is likely to emerge as a seedling or vegetative sprout following a period of quiescence. Weed species also differ with regard to their responses to resource conditions and allelochemicals. Knowledge of these species-specific responses can suggest ways to reduce weed density and impair weed performance through management practices that affect soil fertility and moisture, light interception, and the quantity, quality, and location of crop residues and other organic matter amendments.

Weed Management in Low-External-Input and Organic Farming Systems

289

Biological and ecological data, in conjunction with information obtained from agroecosystem analyses and vegetation monitoring, create the knowledge base from which tactical options for weed management can be evaluated. Enlarging this knowledge base as tactics are tested under field conditions will encourage rapid development of weed management strategies that effectively attack vulnerable points in weed life cycles. TACTICAL OPTIONS As noted earlier, cultivation by machine or by hand is the primary curative approach for managing weeds in farming systems managed without herbicides. It is also the tactical option most often employed when herbicide use is reduced in LEI systems. Important innovations and improvements in cultivation equipment are now being made, particularly in Europe, and the effectiveness of mechanical tactics is increasing. Nevertheless, cultivation used alone as the major line of defense against weeds may still degrade soil, damage crops, and control weeds inconsistently. These shortcomings indicate the need for additional weed management tactics, particularly those that improve the prevention of future weed problems, rather than the short-term cure of existing problems (Jordan, 1996; Bastiaans et al., 2000). Different tactical options affect weeds at different life stages: within the soil seed bank and during germination, seedling emergence, vegetative growth and resource consumption, reproductive growth, and dispersal. By combining tactics that suppress weeds at several life stages, long-term regulation of weed populations is likely to be more successful. Moreover, because of the diversity of physiologies, morphologies, and phenologies present in different weed species, combinations of diverse tactics are more likely than narrow strategies to be successful for managing multispecies weed communities. Here, we briefly summarize some of the important non-herbicide options available to LEI and organic farmers. Tillage Practices Tillage practices affect weed seed placement within the soil profile and physical characteristics of the soil (Cousens and Moss, 1990; Mohler, 1993). Weed seeds initially deposited on the soil surface are buried deepest by moldboard plowing, but this technique also brings back to the soil surface older, dormant seeds that previously had been buried deeply. In contrast, minimum- and zero-tillage practices retain weed seeds on or close to the soil surface. The temperature, atmospheric, and light conditions to which weed seeds are exposed are strongly altered when plowing disturbs soil; these conditions are more constant when tillage is minimized. In general, moderate to severe soil disturbances enhance the germination of most weed species (Mohler, 2001a). Tillage effects on weed seedling density are determined by the combined impacts of weed seed placement within the soil profile, seed dormancy status at the time of burial, seed survival at different depths, germination responses to soil disturbance, and the ability of seedlings to emerge from different depths (Baskin and Baskin, 1989; Mohler, 1993). In general, weed seed survival decreases with proximity to the soil surface, whereas the probability that a newly germinated seedling will emerge

290

Liebman, Bastiaans and Baumann

from the soil increases with the seed’s proximity to the soil surface, particularly for small-seeded species in disturbed soil (Mohler, 1993). Across species, the maximum seed burial depth from which a weed seedling can emerge successfully is correlated with seed mass: species with heavier seeds can emerge from greater depths in the soil (Roberts et al., 1982; Benvenuti et al., 2001). Minimum-tillage systems can be either advantageous or disadvantageous for weed management, depending on whether soil conditions are adequate for weed germination and emergence, and whether new seeds are added to the soil seed bank (Mohler, 1993, 2001b). If weed seeds germinate, but replenishment of the seed bank is strongly restricted through management practices, annual weed populations are likely to decline more rapidly in minimum-tillage conditions than in plowed soil. Alternatively, if weed emergence and reproduction are relatively unfettered and seeds remain near the soil surface, weed populations will increase more rapidly under minimum-tillage conditions than in plowed soil. In many LEI and organic farming systems, flexible tillage strategies may be the best choice for weed management (Mohler, 2001a). After seasons in which weed seed production is strongly suppressed, fewer weeds are likely to be present in the succeeding crop if minimum-tillage techniques are used. In contrast, after seasons in which weed seed production is not effectively suppressed, moldboard plowing can be useful for burying seeds that would otherwise germinate in the succeeding crop. A certain percentage of these deeply buried seeds will die before being returned to the soil surface by subsequent tillage. Cussans and Moss (1982) modeled the effects of tillage on the population dynamics of blackgrass (Alopecurus myosuroides Huds.) and found that deep plowing at least every fifth year was required to counteract increases in blackgrass densities that occurred in years of zero-tillage. Their predictions closely match actual farmer practices for producing winter wheat in the UK (Cousens and Mortimer, 1995, p. 198). Similarly, Sakai et al. (1999) modeled the effects of different depths of tillage on densities of annual weeds and predicted that weed seedling density would be lowest when deep tillage (40 cm) was interspersed in alternate years with shallow tillage (15 or 20 cm depth). Alternation between tillage systems also varies selection pressures on perennial weeds, which are favored by minimum tillage systems and suppressed by plowing (Mohler, 2001a). Thus, over time, the interspersion of plowing with minimum- or zero-tillage is likely to prevent the build-up of weed species welladapted to a particular tillage regime, as well as offer the farmer opportunities to address weed pressures that change due to variable weather conditions, conflicts with other farm activities, labor availability, and other factors. Combinations of tillage practices are also possible in a single field in a single year through the use of "zone tillage" techniques in which soil is disturbed only in a narrow band where the crop is sown. Although the approach is not yet well developed, zone tillage may be useful for restricting soil disturbance and weed seedling emergence to only a small fraction of the field area, thus allowing cultivation and other direct weed management tactics to be concentrated spatially. For example, Forcella and Lindstrom (1988) and Exner and Thompson (1992) described a ridge tillage system in which early-season soil disturbance is confined to the truncation of soil ridges as crop seeds are sown into the ridges. Inter-row cultivation later in the season kills weeds between ridges and buries weeds emerging with the crop on ridges. Melander and Rasmussen (2000) experimented with a biennial cultivation system in which narrow bands of soil are repeatedly cultivated during and after the production

Weed Management in Low-External-Input and Organic Farming Systems

291

of a cereal crop with the goal of depleting weed seed banks, and then, in a succeeding year, a crop such as sugarbeet is sown without further soil disturbance into the previously cultivated bands. Weeds between the bands can be controlled by harrowing. Approaches such as these offer important opportunities for wedding soil conservation with weed management and could be refined if they receive more attention from researchers and equipment manufacturers. Soil Fertility and Moisture Management Because of their fixed root habit, weeds (and crops) are highly responsive to soil chemical, physical, and biological conditions. Furthermore, because small advantages in early season seedling growth can translate into large differences in size and light interception ability later in the season, early responses to soil conditions can be critical for determining competitive interactions between weeds and crops. Fertilizer application and irrigation are two common farming practices that affect soil resource conditions in ways that can be used for weed management. Using these practices to manipulate soil conditions for weed management depends on creating and exploiting differential responses between weeds and crops such that weeds are inhibited and crop performance is enhanced. However, because weed species differ in their adaptations to soil conditions, it should be expected that reliance on any one soil management regime would result in a change in weed community composition. Thus, a diversity of soil management techniques is needed to manage diverse weed communities. Experiments conducted with synthetic fertilizers indicate that nutrients can be placed where they are available to crops, but less available to weeds (DiTomaso, 1995). For example, Rasmussen et al. (1996) found that banding nitrogen 5 cm below the depth at which barley was sown decreased weed biomass 55% and increased barley grain yield 28%, compared with broadcasting the same quantity of fertilizer on the soil surface. The investigators concluded that fertilizer placement close to the crop row favored the crop at the expense of the weeds. Similarly, injection of liquid pig or cow manure into soil at a depth of 8 to 10 cm reduced weed density and biomass and increased barley yield, relative to surface application of manure followed by shallow (3 to 4 cm depth) incorporation with a cultivator (Rasmussen, 2002). The timing of fertilizer application can also be exploited for weed management. For some weed-crop combinations, crop growth and competitive ability are favored by delayed application of nutrients. Angonin et al. (1996) found that over a range of densities, biomass of ivyleaf speedwell (Veronica hederifolia L.) was reduced by half when N fertilizer was applied at the stem elongation stage of wheat development compared to when it was applied earlier, at the tillering stage. The weed had no effect on wheat N uptake and yield when N fertilizer was applied at stem elongation, whereas it reduced wheat N uptake and yield when fertilizer was applied at tillering. The investigators noted that growth, development, and potential competitive effect of ivyleaf speedwell were concentrated early in the growing season, and that delaying N application reduced the weed's competitive ability while benefiting the crop. For other weed-crop combinations, early pulses of fertilizer may provide greater benefit to crops than later pulses. In experiments with winter wheat infested with downy brome (Bromus tectorum L.), Ball et al. (1996) found that delaying application of N fertilizer until spring had either no effect or a stimulatory effect on weed

292

Liebman, Bastiaans and Baumann

biomass production, and either no effect or a negative effect on wheat yield, compared with N application at planting. Nitrogen fertilizer increased wheat yield without increasing downy brome biomass only when it was applied during the fallow period preceding crop production. Anderson (1991) attributed this result to movement of fertilizer N into lower layers of the soil profile before growth of the crop and weed began, and to differences in their rooting habits: the shallow rooted weed was unable to extract fertilizer N at lower depths, whereas the deeper rooted crop gained access to that source of nutrients. Consideration of this case and that of wheat infested with ivyleaf speedwell leads to the conclusion that the impacts of different times of fertilizer application depend on the specific characteristics of the weed and crop species involved. Animal manures offer important opportunities for recycling nutrients and improving soil physical, chemical, and biological properties, but they can also carry weed seeds into fields where they had not previously been present (Dastgheib, 1989; Mt. Pleasant and Schlather, 1994). Aerobic composting is one way to address this problem, although its effectiveness can be species-specific. Eghball and Lesoing (2000) reported that aerobic composting of dairy manure and beef cattle feedlot manure was entirely effective in killing seeds of common cocklebur (Xanthium strumarium L.), giant foxtail (Setaria faberi Herrm.), ivyleaf morningglory (Ipomoea hederacea (L.) Jacq.), redroot pigweed (Amaranthus retroflexus L.), and shattercane (Sorghum bicolor (L.) Moench), but killed only 83 to 96% of velvetleaf (Abutilon theophrasti Medicus) seeds. Both temperature elevation and exposure to chemical products of decomposition, such acetic acid, are responsible for weed seed mortality during the composting process (Shirilipour and McConnell, 1991; Grundy et al., 1998; Ozores-Hampton et al., 1999). Turning and mixing compost is necessary to prevent weed seeds on the outer surface of piles from escaping the thermal and chemical conditions responsible for seed mortality. Irrigation water can serve as a vector for weed seeds (Kelley and Bruns, 1975; Wilson, 1980), but this problem can be reduced by grazing, mowing, or burning ditch banks to keep them free of weeds and by screening irrigation water to remove weed seeds before they enter crop fields (Mohler, 2001b). Certain irrigation techniques can deliver water to crops while minimizing its availability to weeds. Grattan et al. (1988) compared three irrigation methods for producing tomato under nearly rain-free conditions and found that subsurface drip irrigation greatly reduced weed growth and increased tomato production, compared to furrow and sprinkler irrigation. The former method concentrated moisture in the crop root zone, whereas the latter two methods dispersed water across the field surface, making moisture more accessible to weeds. Irrigation timing can strongly affect the abundance and growth of weeds in rice fields. Moody and Drost (1983) reported that the density and biomass of Chinese sprangletop (Leptochloa chinensis (L.) Nees) were nil when rice was flood irrigated five days after seeding, but increased progressively to 72 plants m-2 and 46 g m-2, respectively, as flooding was delayed until 20 days after planting. While these results suggest that early flood irrigation is desirable for rice production, continued reliance on this approach may create shifts in weed community composition toward adapted weed species. This has occurred in California, where water-seeding of rice and continuous flooding began in the 1920s as a method to control severe infestations of barnyardgrass (Echinochloa crus-galli (L.) Beauv.), which is well adapted to dryseeding, but not early flooding (Seaman, 1983). Over time, however, early watergrass (Echinochloa oryzoides (Ard.) Fritsch) and late watergrass (Echinochloa phyllopogon

Weed Management in Low-External-Input and Organic Farming Systems

293

(Stapf) Koss.) have become important weeds and have largely replaced barnyardgrass. The former two species have large seeds, which allow germination and emergence through flood water up to 30 cm deep; barnyardgrass has smaller seeds and is unable to emerge through deep water (Barrett, 1983; Seaman, 1983). As a consequence of this species shift, rice growers in California are now advised to consider alternating between water-seeding and dry-seeding practices (Williams et al., 1992). Crop Residue Management and Mulching Practices Crop residue management is critical for conserving soil and improving soil quality. It can also be used to regulate weed emergence and growth. When incorporated into soil, residues of many crop species, including cereals, crucifers, and legumes, release chemical compounds (allelochemicals) that can inhibit germination and early seedling growth of many plant species (Boydston and Hang, 1995; Weston, 1996). Liebman and Davis (2000) suggested that allelochemicals are likely to be more toxic to smallseeded species (e.g., redroot pigweed) than to large-seeded species (e.g., soybean) or to transplanted seedlings for three reasons. First, small-seeded species tend to germinate close to the soil surface, where concentrations of residue-derived phytotoxins may be highest. Second, small-seeded species tend to have greater amounts of root length per unit of root mass and thus proportionally greater amounts of absorptive surface area through which phytotoxins may enter. Third, compared with large-seeded species and transplants, small-seeded species have fewer reserves with which to tolerate or detoxify chemical stress agents. Because the seeds and vegetative propagules of most agronomic crops are considerably larger than those of the weeds that infest them (Mohler, 1996), allelopathic crops might be exploited for their differential effects on crops and weeds in a range of cropping systems. For example, Dyck and Liebman (1994) reported that incorporation of crimson clover residue into soil reduced the emergence rate, density, and biomass production of common lambsquarters (Chenopodium album L.), a smallseeded weed, while having negligible effects on the emergence and growth of sweet corn, a large-seeded crop. Similarly, Boydston and Hang (1995) found that rapeseed residue incorporated into soil before planting potato reduced weed density 73-85% and reduced weed biomass 50-96%, while raising potato tuber yield 10-18%, compared with a fallow treatment. Boydston and Hang (1995) suggested that isothiocyanates were responsible for the weed suppression, and that the residue’s positive effect on potato growth may have been due to improved crop nutrition, suppression of soil-borne pathogens, as well as reduced competition from weeds. Crop residues used as surface mulches can suppress weed emergence and growth due to both allelopathic and physical effects (e.g., changes in soil thermal regime) (Teasdale, 1998). Crop residues on the soil surface can also reduce weed densities by physically impeding weed seedling emergence and intercepting light that cues weed germination (Teasdale and Mohler, 2000). Differential effects of residues on crops and weeds might be obtained with either zone tillage techniques that remove residue just over the crop row (Kaspar et al., 1990), or the use of crop seeds with large size, independence from light requirements for germination, and enough vigor to push through several centimeters of mulch material (Liebman and Mohler, 2001). The crop residue/mulch approach has been used successfully for weed management in soybean

294

Liebman, Bastiaans and Baumann

(Liebl et al., 1992), sweet corn (Mohler, 1991), and other crops. Special attention must be directed, however, to the management of surface germinating weeds, such as dandelion (Taraxacum officinale Weber in Wiggers), which can be well-adapted to minimum tillage systems (Liebman et al., 1995). Weed seed density can be substantially reduced before planting a crop by covering soil with polyethylene tarps for several weeks, a process that has been called “solarization” (Stapleton and DeVay, 1995). Solarization kills weed seeds by raising soil temperatures, as well as by altering soil atmospheric, chemical, and microbiological conditions. Incorporation of certain cover crop residues into soil before solarization can increase control of soil-borne plant diseases (Gamliel and Stapleton, 1997), and this combination of practices might also increase weed suppression. The costs of solarization tarp materials, installation, removal, and disposal - are relatively high (US $ 750 to $ 1500 ha-1; Stapleton and DeVay, 1995) and probably can be borne only for high value crops. The development of new thinlayer mulch materials that are temporarily effective for sealing the soil surface, but which are photodegradable or biodegradable, would reduce the consumption of petrochemicals in the manufacture of mulch materials and might reduce costs of solarization and mulch disposal (Stapleton and DeVay, 1995). A variety of materials can be added to fields as mulches to suppress weed seedling recruitment during periods of crop production, as well as to conserve soil moisture (Monks et al. 1997; Schonbeck, 1998). Polyethylene is widely used for this purpose, but a number of lower-cost organic materials also are available. Hay and straw can be used, but may serve as sources of germinable seeds and ultimately plants that compete with crops. The use of composted yard wastes, newspaper, and other organic materials that do not contain viable seeds may avoid these problems. Areas between crop rows not covered with mulch require additional weed control measures. Management Practices to Increase Crop Competitive Ability The ability of crops to yield well in the presence of weeds and to suppress weed establishment, growth, and reproduction can be affected by a number of farming practices that increase the proportion of available resources captured by crops rather than weeds. Plant size early in the growing season strongly affects interspecific interactions. Transplanting rather than direct seeding horticultural crops, such as tomato, provides an initial size advantage that can translate into reductions in the length of the critical period for weed control, in weed biomass production, and in crop yield loss to weed competition (Weaver, 1984; Weaver et al., 1992). For cereal crops, such as wheat and barley, seedling emergence rate and early-season height and mass have been shown to be correlated positively with seed size (Chastain et al., 1995a, 1995b). Kolbe (1980) separated barley seed populations into fractions with lighter or heavier seed weights and showed that stands grown from heavier seeds had higher yield and lower weed cover than stands grown from smaller seeds. Substantial increases in crop competitive ability can be achieved by ensuring that weed emergence occurs after crop emergence (Cousens et al., 1987; Chikoye et al., 1995). Differences between crops and weeds with regard to phenology and response to temperature conditions can be exploited for this purpose through adjustments in tillage practices and planting date. For warm-season crops, such as maize and soybean, delaying planting for several weeks beyond the normal date can allow large

Weed Management in Low-External-Input and Organic Farming Systems

295

numbers of weed seedlings to emerge and be killed before the crop is sown (Gunsolus, 1990; Forcella et al., 1993). The stale seedbed approach, which is widely used in carrot and onion production, involves soil disturbance by tillage several weeks ahead of crop sowing to promote weed emergence; after weeds are killed by flaming or shallow cultivation, the crop is sown with minimal further soil disturbance (Mohler, 2001a). For crops that are better adapted to cool conditions than the weeds that infest them, early planting can give the crop a competitive advantage. Vezina (1992) reported that two warm-season grasses, yellow foxtail (Setaria pumila) and barnyardgrass, had no effect on spring wheat yield when the crop was sown in late April or early May, but reduced wheat yield when the crop was sown later. Crop production often occurs in row widths designed to accommodate cultivation equipment, but where row widths can be narrowed and crops planted in more equidistant arrangements, earlier shading of interrow areas can be achieved and crop competitive ability sometimes can be increased (Teasdale and Frank, 1983; Murdock et al., 1986; Murphy et al., 1996). The effectiveness of narrowing row width for weed suppression appears to be linked to the relative heights of the crop and weed species involved. Weeds that are inherently shorter than the crop are likely to be more susceptible to a narrowing of crop rows, whereas weed species that have the potential for growing taller than the crop are much less likely to be susceptible to row narrowing (Schnieders, 1999; Mohler, 2001c). Reductions in row width are often accompanied by increases in crop population density, but even without a reduction in row width, increases in density generally make crops more competitive against weeds. Species for which this has been demonstrated include maize, pea, peanut, rapeseed, rice, safflower, soybean, and wheat (Mohler, 1996). Increases in crop density raise the quantity of resourceintercepting surfaces (leaf and root surface area) that the crop has early in the season and result in a greater proportion of available resources being captured by the crop, rather than by weeds (Mohler, 2001c). Crop genotypes that have high leaf area expansion and height growth rates early in the season often have superior ability to suppress weed growth (Mohler, 2001c). Nevertheless, on an absolute basis, yields of cultivars with high competitive ability against weeds may be inferior to yields of less competitive cultivars. Bastiaans et al. (1997) compared two rice varieties differing in growth characteristics and found that under weed-free conditions, the shorter variety with slower leaf growth (IR8) produced 57% more grain than a taller, leafier variety (Mahsuri). In mixtures with purple rice (a weed), IR8 still produced considerably more grain than Mahsuri (25 to 48%), despite the fact that the biomass of purple rice in plots sown with Mahsuri was 80% less than in plots sown with IR8. This trade-off between competitive ability and yield potential exists in other crops, although it is not universal (Mohler, 2001c). Where the trade-off is observed, breeding efforts are needed to determine whether high competitive ability and high yield potential can be expressed in a single genotype. Intercropping combines two or more crops whose resource consumption characteristics are physiologically, temporally, or morphologically complementary. By combining crop species that differ in the way they use light, water, and nutrients, intercropping can prevent the crops from fully competing with one another (Vandermeer, 1989). Consequently, intercrops may use a greater share of available resources and produce more yield per unit land area than monocultures of the component species (Willey, 1979, 1990). Greater resource use by intercrops than

296

Liebman, Bastiaans and Baumann

monocultures can, in turn, provide improved opportunities for suppressing weeds through resource competition. For example, Dyke and Barnard (1976) reported that the biomass of quackgrass (Elytrigia repens (L.) Nevski) in barley intercropped with red clover was 29 to 88% less than in monocultured barley. Although Dyke and Barnard (1976) did not measure barley yield, results of experiments conducted by Stewart et al. (1980) and Kunelius et al. (1992) indicate that intercropping red clover with barley has little or no effect on barley yield, at least in humid climates, suggesting that increased weed suppression through barley/red clover intercropping need not come at the cost of a yield penalty. Abraham and Singh (1984) found that a grain sorghum/fodder cowpea intercrop intercepted more light, captured greater quantities of nutrients (N, P, and K), produced higher yields, and contained lower weed densities and less weed biomass compared with monocultured sorghum. Similar patterns of resource use, crop yield, and weed suppression have been observed for intercrops and monocultures of sorghum and pigeonpea (Natarajan and Willey, 1980a, 1980b; Shetty and Rao, 1981). Results of experiments conducted in a range of cropping systems indicate that various management practices which increase crop competitive ability can be combined effectively. Malik et al. (1993) found that increased bean density, reduced interrow distance, and the use of a semi-vining, rather than a bush variety maximized bean seed yield and minimized weed biomass production. Teasdale (1995) reported that when maize density was increased and row spacing was decreased, herbicide use could be reduced 75% without any reduction in maize yield and weed control relative to results obtained from a normal density, normal row spacing treatment receiving standard herbicide rates. Bulson et al. (1997) observed that total crop yield rose and weed biomass decreased as the sowing density of wheat/faba bean intercrops was increased above levels used for normal density monocultures of the component crops. Cropping System Diversification in Time Temporal diversification of cropping systems through use of rotations and cover crops can improve nutrient accretion and retention, soil structure, soil conservation, and suppression of certain insect pests and crop diseases (Smith et al., 1987; Hargrove, 1991; Karlen et al., 1994). Consequently, these practices play key roles on low-external-input and organic farms. Crop rotation and cover cropping can also contribute to weed management due to effects on the availability of regeneration niches and growth resources used by weeds (Liebman and Dyck, 1993; Teasdale, 1998; Liebman and Staver, 2001). Several salient points and illustrative examples are noted here. A regeneration niche comprises a species-specific set of environmental conditions required to ensure that a mature plant is replaced by an individual of the next generation (Grubb, 1977). By rotating crops with different planting dates and growth periods, contrasting competitive characteristics, and dissimilar management practices, the regeneration niches of different weed species can be disrupted and increases in densities of key weed species prevented. Blackshaw (1994) reported that downy brome density remained relatively stable when winter wheat was rotated with springsown rapeseed, whereas the weed's density increased rapidly when wheat was grown continuously. Suppression of downy brome was attributed to changes in the herbicide regime and in the timing of tillage practices that reduced seedling survival. Covarelli

Weed Management in Low-External-Input and Organic Farming Systems

297

and Tei (1988) noted that when herbicides were withheld, seed and seedling densities of the summer annual weed barnyardgrass were lower in a cool season/warm season crop rotation (winter wheat/maize) than in a continuous summer crop sequence (maize/maize). Schreiber (1992) found that giant foxtail seedling density tended to be greatest in continuous maize, intermediate in a two-year maize/soybean rotation, and lowest in a three-year maize/soybean/winter wheat rotation. Suppression of giant foxtail in the winter wheat phase of the rotation was attributed to allelopathic effects of wheat straw, but also may have been influenced by phenological differences between the weed and the wheat crop. As the intensity of chemical weed control is lessened, the importance of disrupting regeneration niches through rotation increases. Kegode et al. (1999), for example, found that weed seed production could be reduced to low levels in continuous maize and maize-soybean sequences when standard herbicide doses were used, whereas weed seed production increased greatly in these simple cropping systems when herbicide use was minimized. In contrast, weed seed production was low to moderate in rotations that included maize, soybean, wheat, and alfalfa, regardless of the intensity of herbicide inputs. Rotation of perennial forage crops, such as alfalfa, with annual crops, such as wheat or maize, can reduce densities of annual weeds in the annual crops, though it may increase the density of perennial weeds and certain annual species adapted for survival and reproduction in forage crops (Entz et al., 1995; Clay and Aguilar, 1998; Ominski et al., 1999). Mowing and grazing regimes have a large impact on weed dynamics during forage phases of a rotation sequence and may have to be adjusted to optimize both forage quality and weed suppression. Pino et al. (1998) used field trials and a matrix model to study population dynamics of broadleaf dock (Rumex obtusifolius L.) in wheat/alfalfa rotations. Model projections indicated that early harvest (starting 1 April) of alfalfa reduced broadleaf dock density, whereas later harvest (starting 25 April) could allow density of the weed to increase. Cover crops that occupy a field before or after periods of 'main crop' production suppress weed growth and the production of weed seeds and vegetative propagules that affect subsequent crops. McLenaghen et al. (1996) compared five different cover crops and a fallow control treatment and found that weed growth was inversely proportional to the ground cover produced by the cover crops. Teasdale and Daughtry (1993) reported that live hairy vetch reduced weed density 70-78% and reduced weed biomass 52-70% compared to a fallow treatment. Measurements made after vetch had completed the majority of its growth showed that an average of 87% of sites beneath the cover crop received