1Department of Plant Science and ²Department of Botany

University of Manitoba

Winnipeg, Manitoba, Canada R3T 2N2

m_entz@umanitoba.ca

The ability of forages to suppress weed growth may provide a viable alternative to chemical weed control and allow crop producers to reduce herbicide inputs. Two projects were conducted: 1) A survey was conducted in western Canada in 1993 and 1994 to investigate weed populations in commercial cereal fields that had been preceded by either alfalfa hay or cereal grain crops. A total of 117 fields were surveyed; approximately half from each field type. Naturally occurring populations of wild oat, Canada thistle, wild mustard and cleavers were lower in cereal fields that had previously contained alfalfa than in cereal fields that had been preceded by a number of annual crops. Lower field uniformity values for Canada thistle and wild oat indicated that these weeds were also more Apatchy@ in the alfalfa-containing rotations. Population differences between field types were nonsignificant for red root pigweed , common lambsquarters and wild buckwheat, while populations of dandelion and stinkweed were greater in cereals followed alfalfa vs. cereal following cereal fields. No consistent effect of field type on green foxtail populations was observed.

Project 2: Field studies were conducted in 1994/95 and 1995/96 to investigate the effect of single year forage crops on the density and community composition of annual weeds, both in the seedbank and in a following field pea crop. All forage systems were at least as effective as the sprayed wheat control in suppressing wild oat, however, effects on other weeds, especially broadleafs, were variable. Biennial crops provided the best early season weed control, while long-season systems such winter triticale and a triticale intercrop provided the best late season weed control. Forages shifted the weed community composition away from wild oat and green foxtail to a similar or greater extent than herbicide-treated wheat. Forage systems that did not provide season-long crop competition tended to have more broadleaf weeds.

Herbicide resistant weeds, the need to reduce crop production costs as well as public and environmental concerns regarding pesticide use are reasons why complete reliance on herbicide use in weed control should be decreased (Burnside, 1993; Lotz et al., 1995; Roush et al., 1990; Swanton and Weise, 1991; Wyse, 1994).

Perennial forages, most notably alfalfa, have been cited by many authors as crops that are well suited to compete with weeds (Stevens, 1846 - cited in Donald, 1990; Hunter et al., 1990; Pavlychenko, 1942; Siemens, 1963; Todd and Derksen, 1986). Peters and Linscott (1988) state that summer annual weeds seldom infest mature, vigorous and densely growing alfalfa stands. Hodgson (1958) found that alfalfa, cut for hay twice a year, reduced the number of Canada thistle (Cirsium arvense (L.) Scop.) shoots to 1% of the original stand in a three year period. Similarly, Derscheid et al. (1961) reported almost complete elimination of the thistle population on land heavily infested with this weed after three years of alfalfa hay. In a study that investigated the competition between selected crops and field bindweed (Convolvulus arvensis L.), Stahler (1948) remarks upon the "inherent ability" of alfalfa to successfully compete for sunlight and soil moisture.

Danish researchers found that the average frequency of common weed species was different in alfalfa and perennial grass stands than in annual crops such as barley (Hordeum vulgare L.), field beans (Phaseolus vulgaris L.) and canola (Brassica napus L.) (Haas and Streibig, 1982). They also noted that wild buckwheat (Polygonum convolvulus L.) and common lambs quarters (Chenopodium album L.) almost disappeared in biennial and perennial crops. Pallas (1994 - cited in Stupnicka-Rodzynkiewicz, 1996) reported that inclusion of alfalfa in rotation prevented subsequent crops from being infested with weeds typically found in a cereal rotation. However, Thomas et al. (1996) found that rotations that included hay crops had only slightly different weed communities than did rotations without hay. In a western Canadian survey, 84% of producers indicated that the inclusion of alfalfa in crop rotations reduced weed populations in subsequent annual crops (Entz et al., 1995). In particular, producers noted good suppression of wild oat (Avena fatua L.), Canada thistle, green foxtail (Setaria veridis (L.) Beauv.) and wild mustard (Brassica kaber (DC.) L. C. Wheeler).

The weed control benefits of innovative annual forage systems like spring-sown winter cereals and winter/spring cereal intercrops (Baron et al. 1992) have not been evaluated. The objectives of this second study were: 1) to determine how effectively single year forages suppressed annual summer weeds, and 2) to compare annual forage systems with the conventional weed control system (grain crop with herbicides) on the basis of weed density and community composition in a subsequent pea crop.

.

Study 1 A survey was conducted throughout Manitoba, Canada in early June of 1993 and 1994 to determine weed populations in commercial cereal fields that had been preceded either by a cereal crop (>cereal following cereal fields=) or by an alfalfa or alfalfa/grass crop (>cereal following alfalfa fields=). The spring-seeded cereal test crops included mostly wheat, however some barley or oat (Avena sativa L.) crops were also sampled

Survey fields were equally distributed throughout the cropping zone of Manitoba. The sampling technique was based on the Ainverted W@ method used by others to survey weed infestations in agricultural fields (Thomas, 1985). All surveys were conducted by members of the Manitoba Weed Supervisors Association, a group weed specialists with considerable experience using the Thomas weed surveying method.

Study 2 Field experiments were conducted at the University of Manitoba field research station at Carman, Manitoba from 1994 to 1996. Indigenous summer annual weed populations in the trial area were comprised predominantly of green foxtail, red root pigweed, common lambsquarters, wild buckwheat, wild mustard and green smartweed. Wild oat seeds were broadcast onto the plot area prior to initiation of the study. The two-year field trials were conducted twice; once in 1994/95 (trial 1) and again in 1995/96 (trial 2). In the first year, annual forage "suppressor" crops were seeded and grown in a herbicide-free environment. In year two, a field pea test crop was grown both with and without in-crop herbicides. In this way, the economic weed control of the suppressor treatments could be determined. The nine treatments for trial 1 were: wheat (cv. Katepwa) sprayed with grass and broadleaf herbicide (sprayed wheat), harvested for grain; wheat sprayed with broadleaf herbicide only (partially sprayed wheat), harvested for grain; wheat, unsprayed (unsprayed wheat), harvested for grain; winter triticale (Triticosecale cv. Pika), simulation grazed; spring triticale (cv. Banjo), cut for silage; winter and spring triticale intercrop (IC), cut for silage then simulation grazed; sorghum sudangrass (common seed), cut for hay; non-dormant alfalfa (cv. Nitro), cut for hay; and weed fallow, cut for silage. Two additional treatments were added in trial 2: a sweet clover (Melilotus officianales L. cv. Norgold)/winter triticale doublecrop (DC), cut for hay then simulation grazed; and fall rye (cv. Prima), harvested for grain. Initial weeds densities in the forage crop year were determined on May 31 in each year. Successive weed density measurements were taken immediately prior to the respective forage crop harvests at each harvest date. Weed seedlings were identified, counted and their height was measured.

Study 1.

In 1993, total field density (the sum total of weed species density values) was 48 plants/m² for cereal following alfalfa fields and 106 plants/m² for cereal following cereal fields (Table 1). Cereal following alfalfa fields contained 27 different species while cereal after cereal fields contained 31 species. In 1994, total field density was 81 plants/m² for cereal following alfalfa fields and 128 plants/m² for cereal following cereal fields (Table 2). Cereal following alfalfa fields contained 41 species while cereal following cereal fields contained 34 species. Higher total field densities and greater species diversity in 1994 is reflective of the year to year variability often expressed by weed populations due to annual fluctuations in environmental conditions (Derksen et al., 1993; Chepil, 1946; and Radosevich et al., 1997).

Despite some year to year variation, results within field types were similar between years. Weed species density values for cereal following alfalfa fields ranged from a minimum of 0.01 povertyweed plants/m² to a maximum of 8 wild mustard plants/m² in 1993 and from 0.01 cranesbill plants/m² to 21 green foxtail plants/m² in 1994. The greater range of densities observed in cereal following alfalfa fields in 1994 was due to a large increase in green foxtail density. When green foxtail was excluded from the analysis, the range of densities (0.01 - 9 plants/m²) were similar in both years. Density values recorded for cereal following cereal fields were also similar for both years and ranged from 0.02 sweet clover plants/m² to 27 wild oat plants/m² in 1993 and from 0.03 pineapple weed plants/m² to 29 green foxtail plants/m² in 1994. Results obtained for cereal following cereal yields in the present study were similar to previous surveys conducted in Manitoba (Thomas and Donaghy, 1991 and Thomas et al., 1997).

Relative abundance values, which are calculated to summarize the contribution of individual species to the weed community (Thomas, 1985), were also consistent from year to year for each field type. The ten most abundant weed species accounted for more than 80% of the total relative abundance value (the summed total of all weed species relative abundance values) for both field types and in both years. Of the 10 most abundant weeds between field types for 1993, seven species (mainly annual dicots) were similar to both data sets (Table 2). The most striking differences were: 1) the absence of green foxtail and wild oat as the first and second most abundant species in cereal following alfalfa fields and 2) the absence of volunteer alfalfa, dandelion, and stinkweed in cereal following cereal fields. In 1994, seven of the ten most abundant species were common to both field types (Table 3). Once again, stinkweed, volunteer alfalfa and dandelion were present among the ten most abundant weed species in cereal following alfalfa fields and absent from this category for cereal following cereal fields . As was the case in 1993, wild oat was absent from the top ten weeds in cereal following alfalfa fields in 1994. However, green foxtail was the most abundant weed in both field types in 1994.

Annual Grasses

Based on relative abundance values, wild oat and green foxtail were the only annual grass species of major importance in this survey (Table 1 and 2). Of the two species, the results obtained for wild oat were most consistent. Wild oat densities in cereal following alfalfa fields were 1 and 0.4 plants/m² in 1993 and 1994, respectively, compared with 27 and 13 plants/m², respectively for cereal following cereal fields (differences significant at p < .01 for both years). It is apparent from these data that alfalfa effectively suppresses wild oat growth and seed return as has been observed by Siemens (1963). This suppressive effect may be due the highly competitive root and shoot growth of alfalfa. Cutting regime may also influence wild oat population dynamics. Because wild oat seed can shatter before a cereal crop is harvested (Thurston, 1966; Chepil, 1946), annual replenishment of the seed bank will occur each year in an annual crop rotation. However, the two hay cuts harvested per growing season from an alfalfa hay stand hay are known to reduce seed set and subsequent seed return (Schoofs, 1997). Effects of alfalfa in rotation were less consistent between years for green foxtail than for wild oat.

Annual dicots of major significance in this survey included wild buckwheat, wild mustard, annual smartweed, red root pigweed, and common lambsquarters. These species were consistently among the top ten most abundant species for both field types and survey years. This confirms that, in addition to green foxtail and wild oat, these are the most important weed species in Manitoba (Table 1 and 2). Two of these weeds, wild mustard and annual smartweed, plus a less abundant weed, cleavers, were strongly influenced by inclusion of alfalfa in rotation. For example, wild mustard density in cereal following alfalfa fields was 8 and 7 plants/m² in 1993 and 1994, respectively, compared with 11 and 23 plants/m², respectively in cereal following cereal fields (p < .10 in 1993 and p < .05 in 1994). Frequency of wild mustard was lower in cereal following alfalfa in 1993 (Table 1), although no differences between field type were observed in 1994 (Table 2). Based on frequency observations, it can be concluded that while populations of wild mustard are reduced by alfalfa, they are not eliminated. In 1994, similar trends were exhibited for annual smartweed with significantly fewer plants (p < 0.1 ) present in cereal following alfalfa fields. Cleavers was the third annual dicot whose density was significantly reduced by the inclusion of alfalfa in rotation. This plant is gaining prominence throughout the Canadian prairies as an important weed in canola fields (Anonymous, 1997). Cleavers was ranked as the ninth most abundant species in cereal following cereal fields in 1993 and 1994, compared with relative abundance rankings of 11^th and 17^th, for cereal following alfalfa fields. Frequency of cleavers was lower in cereal following alfalfa fields in both years (Table 1 and 2), while cleaver density was significantly lower (p < .05) in cereal following alfalfa than cereal following cereal fields in 1994. Common lambsquarters, wild buckwheat, and red root pigweed were less affected by the inclusion of alfalfa in crop rotation than the former three species.

The present study supports previous observations of the ability of alfalfa to suppress Canada thistle (Derscheid, 1961; Hodgson, 1958; Stahler, 1948). In 1993, Canada thistle shoot density was significantly lower in cereal following alfalfa vs. cereal following cereal fields (p < .01), while a similar but non-significant trend was observed in 1994 (Table 2 and 3). Alfalfa in rotation was also found to reduce relative abundance of Canada thistle relative to cereal following cereal rotations. Frequency of occurrence for Canada thistle was lower for cereal following alfalfa fields in 1993 (54 vs. 7%), but no differences were observed in 1994 (Table 3). Higher frequency of Canada thistle in 1994 may be attributed to wet conditions the previous summer (Table 4), which may have stimulated Canada thistle seed germination and enhanced vegetative plant growth in the late summer (Moore, 1975). Alfalfa also suppressed perennial sow thistle as density values were significantly lower in cereal following alfalfa fields than in cereal following cereal fields in both years (p < 0.1). As was the case with Canada thistle, the relative abundance and frequency of sow thistle was higher in 1994 than 1993 for both field types, with the greatest increase observed in cereal following alfalfa fields (Table 1 and 2).

Alfalfa in rotation resulted in significant (p < 0.01) increases in dandelion density (4 vs. 1 plant/m^-2), as well as higher dandelion frequency relative to cereal following cereal rotations (Table 1 and 2). Furthermore, dandelion was the fifth and seventh most abundant weed in 1993 and 1994 in cereal following alfalfa fields, compared with 21^st and 13^th most abundant weed in 1994 in cereal following cereal rotations, respectively. It was interesting to note that dandelion population demographics remained relatively stable over years. Therefore, while the thistle species considered in this study had higher densities and levels of occurrence after the wet season of 1993, dandelion appears to have been unaffected.

Greater abundance of dandelion after alfalfa may be due to its prostrate growth habit, which may make it possible for the species to avoid defoliation during mechanical hay harvesting. The germination pattern of dandelion may also be involved. Because dandelion does not exhibit any regular or marked periodicity of germination over a growing season (Chepil, 1946), the potential for successful germination is constant throughout the year. Therefore, the opportunity exists for dandelion to establish itself during the period after a hay cut when much of the soil surface is exposed, or in the fall when alfalfa is dormant. Champness (1949) concluded that dandelion populations could not withstand the competition from an alfalfa crop that was neither grazed nor cut for hay during the growing season. Legere (1993) observed greater dandelion populations in a rotation containing red clover (Trifolium repens L.) and attributed the increase to a more stable environment created by the absence of tillage.

An additional reason for high dandelion populations after alfalfa in our study may be weed invasion due to poor alfalfa stand health. In the 1994 survey, 75% of producers indicated that deteriorating alfalfa vigor was the motivating factor for terminating the stand. Reason for stand termination was not recorded in 1993. Entz et al. (1995) reported that the strategy of most forage producers was to maximize forage stand life and rotate forages only when necessary due to weed invasion and declining hay yields. It is therefore conceivable that the higher presence of dandelion in the cereal following alfalfa fields may partially be was an artifact of poor stand health and that dandelion populations would not be as great if alfalfa stands were more intensely managed to optimize plant vigor and taken out of production prior to severe weed invasion.

Quack grass is another species whose population demographics varied substantially from year to year. Cereal following alfalfa fields contained fewer plants/m² than cereal following cereal fields in 1993 (Table 1).

Winter annual weed species germinate in the fall when alfalfa plants are dormant and resume growth early in the spring prior to the resumption of alfalfa growth. Hence, these weeds can become established in healthy alfalfa stands (Peters and Linscott, 1988). Data on stinkweed populations in the present survey support this conclusion (Table 1 and 2). Stinkweed densities in cereal following alfalfa fields in 1993 and 1994 were 3 and 10 plants/m², respectively, compared to 0.5 and 3.5 plants/m² in cereal following cereal fields; differences were significant in 1994 only.

Uniformity measurements provide an indication of weed patchiness. Results of the present study indicate that while wild oat, Canada and sow thistle, and cleavers were much less uniform (i.e., more patchy) in fields following alfalfa compared with annual rotations, uniformity of dandelion, stinkweed and volunteer alfalfa was greater in cereal following alfalfa vs. cereal following cereal rotations (Table 1 and 2). These observations have several implications. First, alfalfa in a cropping system may facilitate patch-based control measures (such as spot-spraying) for certain weeds (e.g. wild oat), while reducing this option for other weeds (e.g. dandelion). Second, greater patchiness for some weeds after alfalfa may provide clues to the mechanisms of weed suppression with alfalfa. For example, greater patchiness of wild oat after alfalfa suggests that the alfalfa hay crop may have reduced movement out of an original patch, or resulted in a decreased wild oat seed bank in areas outside the dense Amother@ patches.

Study 2

All forage treatments controlled wild oat at least as well as the sprayed wheat control treatment, and winter triticale and the triticale IC had significantly fewer wild oat seedlings than most other treatments (Table 3). Fall rye, which had among the highest weed seedbank density values (data not shown), had very low green foxtail recruitment in the pea crop. Similar effects of fall rye were observed for field emergence of red root pigweed, lambsquarters, and to a lesser extent, wild buckwheat (Table 3). This observation was attributed to allelopathy, which is well documented for rye crops (Weston 1996). Across all weed species, the two forage treatments that performed as well or better than the sprayed wheat control were winter triticale and sorghum-sudangrass (Table 3). All other forage systems displayed a weakness for at least one weed species, usually broadleafs. The treatment which provided no crop competition, namely weed fallow, had among the highest broadleaf weed populations in both the seedbank and the pea test crop. Based on these observations, winter triticale and sorghum-sudangrass were the best weed suppressor crops. The remaining treatments, while providing good control of wild oat and green foxtail (except alfalfa), resulted in higher broadleaf weed populations than the sprayed wheat control.

Results of study 1 showed that alfalfa has had a strong impact on weed species composition and abundance. From data obtained in this study, it is possible to categorize weed species into 4 groups consisting of: 1) wild oat, wild mustard, Canada thistle, sow thistle, and cleavers, whose populations were lower in cereal following alfalfa fields thus indicating greater suppression by alfalfa than by annual crop production practices; 2) wild buckwheat, redroot pigweed and lambsquarters whose populations were similar for both field types, 3) dandelion and stinkweed whose populations increased in an alfalfa stand due to conditions more conducive to growth than in a cereal rotation; and 4) green foxtail for which the data is inconclusive due to large year to year variations.

These results have several implications for weed management. First, inclusion of alfalfa in a rotation is an effective alternative to chemical weed control for certain weeds. In particular, alfalfa hay production can be used as a tool to manage herbicide resistant weeds such as wild oat, which has become an issue of great concern in many parts of the world. Second, for those weeds species where no differences between field types were observed, inclusion of alfalfa in rotation can deliver the same degree of suppression as herbicides in an annual production system. Third, the rise in dandelion and stinkweed populations in cereal following alfalfa fields is a concern that needs to be addressed. Finally, effects of alfalfa on weed patchiness in following crops may in some cases facilitate patch-based management strategies.

From an agronomic perspective, the present study illustrates the need to better understand how alfalfa stand management influences weed populations. Factors such as stand age, alfalfa nutrition and cutting regime are important in determining stand health which, in turn, directly affects the ability of alfalfa to compete with weeds. Alfalfa stand termination method (ie., disturbed vs. non-disturbed) is another factor that will likely affect weed populations in succeeding crops.

Study 2 provided information on the effects of selected annual forage systems on population dynamics of certain annual weed species under moderate and high weed populations. All forage systems were at least as effective as herbicides in reducing wild oat seed "rain". The two forage treatments that effectively reduced seed rain in both grass and broadleaf weeds were winter triticale and sorghum-sudangrass. All forage treatments were at least as effective as herbicide-treated wheat in shifting species dominance away from wild oat, and in some cases also green foxtail. By reducing the population and community dominance of wild oat in the absence of herbicides, these forage systems can play an important role in managing herbicide resistant wild oat populations. Certain forage systems, while controlling wild oat, selected for broadleaf weeds in the community. Annual forages did not eliminate the need for herbicides in a following pea crop, but did contribute to higher pea yields.

The most effective forage systems were those which provided weed competition for the entire growing season. Winter triticale had this attribute; it produced relatively high levels of dry matter from late May until freeze-up in October. Alfalfa, though a long-season crop, provided poor control of broadleaf weeds and green foxtail. Another important attribute of successful systems was rapid early season growth. The clover/triticale DC provided the best early season weed competition of any forage system tested, however a niche was created for weeds after clover harvest. The success of the sorghum-sudangrass crop was attributed to delayed seeding, which is a well-known cultural weed control method, and the fact that it was better able to compete with C4 weeds like green foxtail and red root pigweed. Based on these observations, it was concluded that the ideal system should combine the early season vigour of a biennial crop, the continuous competition of a long-season crop, and the intense mid-summer competition of a warm season crop. It is not clear whether the level of weed control achieved by simulation grazing would be similar to actual grazing. This requires further investigation.

The authors gratefully acknowledge the assistance of the Manitoba Weed Supervisors Association,

who carried out the field surveys. Thanks also to the Soils and Crops Branch of Manitoba Agriculture for assistance in organizing the survey. Funding for this project was provided by a scholarship to P.D. Ominski from the Natural Sciences and Engineering Research Council of Canada, and a scholarship to A. Schoofs from the University of Manitoba. A project grant from the Canada-Manitoba Agreement on Agricultural Sustainability is also gratefully acknowledged..