Weed Science 2010 58:278–288
Glyphosate-Resistant Cropping Systems in Ontario: Multivariate and Nominal Trait-Based Weed Community Structure Robert H. Gulden, Peter H. Sikkema, Allan S. Hamill, Franc¸ois J. Tardif, and Clarence J. Swanton* Glyphosate-resistant (GR) cropping systems are popular and used extensively by producers. However, the long-term impacts of heavy reliance of this technology on weed community structure are not known. Five fully phased field experiments (two no-tillage and three conventional tillage) were established at four locations in southwestern Ontario where the effects of herbicide system (glyphosate or conventional) in corn and soybean and crop rotation (corn–soybean or corn–soybean–winter wheat) on midseason weed communities were examined. Multivariate analysis on data over the last 3 yr of the 6-yr experiment showed that weed communities were distinctly different among the treatments within each experiment. At several locations, midseason weed communities were more similar in corn and soybean treated with glyphosate compared to the same crops treated with conventional herbicides, reflecting the continuous application of the same selection pressure in both crops. Analysis of trait-densities revealed an increase in species with late initiation of seedling recruitment at the expense of weed species with medium time of initiation of seedling recruitment rather than early recruiting species. Increases in perennial species, species with a short interval between recruitment and anthesis, and wind-dispersed species were also observed. Trait-density–based analysis of the weed community was an effective method for reducing the complexity of divergent weed communities that enabled direct quantitative comparison of the herbicideinduced effects on these weed communities. Nomenclature: Glyphosate; corn, Zea mays L. ZEAMX; soybean, Glycine max (L.) Merr. GLXMA; winter wheat, Triticum aestivum L. TRZAW. Key words: community structure, glyphosate-resistant, canonical discriminant analysis, corn, multivariate analysis, soybean, trait-based analysis, winter wheat.
In Canada, one of the first countries to commercialize herbicide-resistant crops, glyphosate-resistant (GR) soybean and GR corn currently make up . 60% (Anonymous 2004) and about 30% (G. Stewart, personal communication) of the total annual acreage of these crops, respectively. In Ontario, where corn and soybean are grown in rotation, the continuous use of GR technology is possible. Glyphosate-resistant cropping systems are popular because they offer many advantages over conventional herbicide systems (Sikkema and Soltani 2007). In Ontario, GR cropping systems also represent a paradigm shift in herbicide use patterns. There, traditional weed control in these crops consisted of preemergence selective, soil-residual herbicides. In contrast, glyphosate, a nonselective, nonresidual post-emergence herbicide, may be applied between two and four times in a growing season. Weed management practices strongly affect the structure of the midseason weed community with in-crop management typically being most influential (Doucet et al. 1999). Similarly, Fried et al. (2008) showed that anthropogenic factors, particularly herbicides were the main factors driving weed communities in farm fields and herbicide properties (e.g., mode of action, time of application) also present significant filters for the weed community (Fried et al. 2009). Crop rotation and growth habit also affect weed community assembly (Swanton et al. 1999, 2006). In Ontario, winter wheat is often grown in rotation with corn and soybean. The winter growth habit and different herbicide chemistry used in DOI: 10.1614/WS-D-09-00089.1 * First author: Assistant Professor, Department of Plant Science, 222 Agriculture Building, 66 Dafoe Road, University of Manitoba, Winnipeg, MB R3T 2N2, Canada; second, fourth, and fifth authors: Associate Professor, Associate Professor, and Professor, Department of Plant Agriculture, Crop Science Building, University of Guelph, 50 Stone Road E., Guelph, ON N1G 2W1, Canada; third author: Research Scientist (retired), Agriculture and AgriFood Canada, Harrow, 2585 Country Road 20, ON N0R 1G0, Canada. Corresponding author’s E-mail:
[email protected]
278
N
Weed Science 58, July–September 2010
winter wheat compared to corn and soybean resulted in distinctly different weed communities in a study by Swanton et al. (2006). Over three complete cycles of the rotation, weed communities were more influenced by management factors during the early years of the study, while being influenced more by crop during the later years of this experiment. In 2000, Shaner (2000) predicted that high use intensities of GR cropping systems would result in changes in the weed community with a shift towards species that are not controlled well by this herbicide. This prediction was based on observations following the introduction and high use of acetolactate synthase–inhibiting herbicides in the early 1980s (Mayo et al. 1995; Sweat et al. 1998). Species with natural tolerance to glyphosate include Ipomoea spp. (Baucom and Mauricio 2004; Marshall et al. 2000), Asiatic dayflower (Commelina communis L.), birdsfoot trefoil (Lotus corniculatus L.), Chinese foldwing [Dicliptera chinensis (L.) Juss], common lambsquarters (Chenopodium album L.), Benghal dayflower (Commelina benghalensis L.) (Nandula et al. 2005), bermudagrass (Cynodon dactylon L.) (Bryson and Wills 1985), field bindweed (Convolvulus arvensis L.) (Duncan and Weller 1987), common waterhemp (Amaranthus rudis J.D. Sauer) (Patzoldt et al. 2002), nutsedge (Cyperus spp.), Canada fleabane (Conyza canadensis L.), and hemp sesbania [Sesbania exaltata (P. Mill.) McVaugh] (Shaner 2000). Previous studies involving GR crops have documented weed shifts. Harker et al. (2005) examined the effects of different use intensities of GR canola (Brassica napus L.) and GR wheat in rotation on quantitative and qualitative weed community composition at several locations in western Canada. After 4 yr of no to continuous use of GR crops, canonical discriminant analysis revealed some separation of the weed communities among the different treatments. Weeds including wild oat (Avena fatua L.), green foxtail (Setaria viridis L.), redroot pigweed (Amaranthus retroflexus L.), sowthistle (Sonchus spp.), and prostrate knotweed (Polygonum aviculare L.) tended to be more associated with the
conventional herbicide system compared to in-crop use of glyphosate. Westra et al. (2008) observed a significant increase in the common lambsquarters population in a GR cropping system compared to conventional herbicides. Similarly, field scale evaluations in the United Kingdom also revealed species shifts in response to herbicide-resistant cropping systems compared to systems in which conventional herbicides were used (Heard et al. 2003). In glufosinate-resistant corn, five species exhibited greater survival rates, including shepherd’s purse [Capsella bursa-pastoris (L.) Medik.], common lambsquarters, black bindweed [Fallopia convolvulus (L.) A. Love], chickweed [Stellaria media (L.) Vill.], and Persian speedwell (Veronica persica Poir.), than when conventional herbicides were used. Glyphosate-resistant cropping systems have been adopted over large geographic regions with unique weed species complements. Unique weed species among fields or geographic regions impose challenges for making comparisons in shifts in community structure in response to management. Recently, some of these challenges have been overcome by reducing species community complexity through grouping species based on common traits (Booth and Swanton 2002; Weiher et al. 1999). For example, Choler (2005) and De Bello et al. (2005) used plant traits to investigate plant community structure along climatic gradients. Trait-based analyses also have been used to describe grassland communities (Lindborg and Eriksson 2005; Louault et al. 2005), the response of plants to grazing (Cingolani et al. 2005; Diaz et al. 2007; Pykala 2004), and the effects of plant–plant competition (Wiegmann and Waller 2006). Redundant traits among species or the definition of broad, meaningful (functional) trait categories enable the investigation of similarities among complex communities. Trait-based analyses of plant communities, however, have only been used sparingly to study agricultural weeds (Booth and Swanton 2002). In the United Kingdom, Storkey (2006) reduced the annual weed community in arable fields to six functional types and related these to the diversity of invertebrates and birds. Trait-based analyses have also been used in agriculture to study plant morphological adaptation to varying grazing intensity (Diaz et al. 1992). Glyphosateresistant cropping systems also appear to select for weed communities with specific traits. There is evidence that the density of species with late recruitment that avoid exposure to glyphosate tends to increase in these cropping systems (Harker et al. 2005; Hilgenfeld et al. 2004a; Marshall et al. 2000; Puricelli and Tuesca 2005). Within the context of the whole weed community, however, these hypotheses have not been tested explicitly. It is also not clear whether GR cropping systems select for other traits for which there may not be an obvious link to glyphosate. The objectives of this study were to compare weed communities in corn and soybean treated with glyphosate or conventional herbicides in rotation with or without winter wheat at five locations with quantitatively and qualitatively different weed communities throughout southwestern Ontario. A multivariate approach was used to examine qualitative differences in weed communities in years 4 to 6 of a series of experiments designed to examine the effects of glyphosate compared to conventional herbicides in corn and soybean production. In addition, a trait-based approach using nominal categorization was used to test hypotheses of expected shifts in weed community structure in response to treatment with glyphosate compared to conventional herbicides.
Materials and Methods
Experimental Design. A detailed description of the locations, site management, and data collection of these experiments can be found in Gulden et al. (2009). In brief, five corn–soybean rotation experiments with and without winter wheat in rotation were established in 2000 in southwestern Ontario under either conventional tillage (CT) (Woodstock, Huron Park, Woodslee) or no-tillage (NT) (Woodstock and Ridgetown). Conventional tillage consisted of primary tillage in fall (moldboard plow or chisel plow) and secondary tillage in spring (disk or light cultivator), and NT consisted of soil disturbance only at planting. Glyphosate-resistant corn and soybean were grown throughout and treated either with preemergence soil-residual conventional herbicides (S-metolachlor/benoxacor and dicamba/atrazine in corn or flumetsulam/S-metholachlor in soybean) or glyphosate (Gulden et al. 2009). In the 3-yr rotation, winter wheat served as a common crop between the systems and was treated with the same herbicides (bromoxynil/MCPA) regardless of the herbicide system used in the previous corn and soybean crops. After harvest, all winter wheat plots were treated with glyphosate. All experiments were fully phased (i.e., each phase of the rotation was present each year) resulting in 10 treatments (2and 3-yr rotations treated with two different herbicide systems) and arranged in a randomized complete block design with four replicates. Individual plots ranged from 60 to 104 m2 in size. Crops were managed using best management practices for the area. A preplant glyphosate ‘‘burndown’’ was applied to all NT experiments. Conventional herbicides were applied after planting but before crop emergence, and in the GR systems one or two in-crop applications of glyphosate were applied (Gulden et al. 2009). Data Collection. In early August (7 to 8 wk after in-crop treatment with glyphosate), the midseason weed community was assessed as described in Gulden et al. (2009). In brief, individual weed species were identified and counted in a minimum of five randomly placed 1-m2 quadrats per experimental unit. Mean weed density (plants m22) of each species was determined for each experimental unit for each year and then subjected to multivariate and trait-density– based analyses. Multivariate Analysis. All species present between 2003 and 2005 were included in the multivariate analysis. Multivariate analysis was conducted within each location to determine whether herbicide system and rotation in corn and soybean resulted in distinctly different weed communities among treatments at each location. To determine which multivariate method was most appropriate, a detrended canonical correspondence analysis was conducted initially and the length of gradient of the eigenvalues was examined (Leps and Smilauer 1999). This indicated redundancy analysis was the most appropriate method. To reduce potential bias of rare species, data were log-transformed and rare species were downweighted during redundancy analysis using CANOCO 4.5 (Leps and Smilauer 1999). Herbicide system and rotation treatment combinations were treated as independent variables. Species densities were treated as dependent variables and years and replications were considered covariables. Monte Carlo permutations were used to test the significance of the first two Gulden et al.: Glyphosate-resistant crops: weed communities
N
279
Table 1. Plant trait groups, functional traits, and categories used to generate the species by trait matrix. Plant trait group 1. Taxonomic characteristics Taxonomic group Life cycle
Classification categories Grass/sedge, broadleaved
Alex and Switzer (1975) Mulligan (1979, 1984) Cavers (1995, 2000, 2005) Alex and Switzer (1975) Mulligan (1979, 1984) Cavers (1995, 2000, 2005)
Annual, biennial, perennial
2. Recruitment biology Recruitment initiation
Primary source
Early (, 20 DAEFa), med (21–40 DAEF), late (. 40 DAEF)
Doll (2007)
3. Flowering characteristics Time to anthesis
Short (# 42 d)
Mulligan (1979, 1984) Cavers (1995, 2000, 2005)
4. Dispersal mechanism Seed dispersal
Wind
Mulligan (1979, 1984) Cavers (1995, 2000, 2005)
5. Predicted to increase Species predicted to increase in GR
Various traits
Duncan and Weller (1987) Krausz et al. (1996) Marshall et al. (2000) Shaner (2000) VanGessel (2001) Patzoldt et al. (2002) King et al. (2004) Owen and Zelaya (2005) Nandula et al. (2005) Hacault and Van Acker (2006)
a
Abbreviation: DAEF, days after emergence of first species.
canonical axes (Økland 2003). The treatment structure was nominal and scaling was focused on intersample distances. Biplots were generated for each analysis using CanoDraw (Leps and Smilauer 1999). Trait-Density–Based Analysis. Prior to trait-based analysis and determination of trait densities within treatments among locations, a trait matrix was constructed. All weed species were classified according to five broad trait categories including taxonomic characteristics, seedling recruitment characteristics, flowering characteristics, dispersal mechanism, and species predicted to increase under glyphosate use (Tables 1 and 2). Traits were chosen primarily based on availability of data. For traits such as recruitment characteristics, there are obvious mechanisms by which herbicide systems filter the weed community. For other traits such as anemochory, filtering mechanisms are less clear and they were evaluated on an exploratory basis. Weeds were categorized based on information available in the literature with primary sources being the Biology of Canadian Weeds series (Cavers 1995, 2000, 2005; Mulligan 1979, 1984), Alex and Switzer (1975) manual of Ontario weeds, and the weed seedling recruitment summary developed by Doll (2007). Based on these categories, a specific binary species by trait matrix was generated for all the species present in each experiment. All trait categories were chosen a priori (Tables 1 and 2). To evaluate the independence of plant traits, the species by trait matrix for each location was subjected to the CLUSTER procedure in SAS.1 Due to the binary nature of the species by trait matrix, the complete method algorithm was chosen to evaluate the relationships among the traits. From this analysis, dendrograms were constructed. 280
N
Weed Science 58, July–September 2010
To determine the relative trait-densities of the weed communities among treatments, each species by trait matrix was multiplied by the treatment by species density matrix generated for multivariate analysis as described by Diaz et al. (1992). Trait-densities were then converted to relative traitdensities by dividing each trait-density by the total density for each experimental unit. All weed species used in the multivariate analysis, including rare species, were incorporated in the trait-based analysis. Relative trait densities did not follow the Gaussian distribution, and therefore data were analyzed using the GLIMMIX procedure in SAS using the log-normal distribution with its identity link function. To reduce the number of high level interactions and allow easy comparison of trends among experiments, the data were analyzed within each crop (corn or soybean) within each experiment with herbicide system and crop rotation as the main factors and fixed effects. Replication and year were considered random effects. For traits with only two categories, trait-densities of only one category needed to be analyzed to determine statistically significant differences among traits. For traits with three categories, however, statistical analysis was conducted on all three categories. To visualize the direction and magnitude of difference among herbicide systems used in corn and soybean, the proportional trait-densities were plotted using the reaction norm approach. Results and Discussion
Treatment Separation in Response to Midseason Weed Communities. Treatment centroids were separated primarily by crop and herbicide system while they were clustered within these by rotation, clearly indicating that crop and herbicide
Table 2. Bayer code, scientific name, common name, and trait categorization of weeds in each experiment. Bayer code
Scientific name
Annual ABUTH AMARE AMASS AMBEL ATXPA CHEAL DIGSA ECHCG ERIAN ERICA HIBTR PANCA PANDI POLCO POLLA POLPE POROL SETFA SETLU SETVI SINAR SOLPT SONOL STEME ZEAMX
Abutilon theophrasti Medik. Amaranthus retroflexus L. Amaranthus species Ambrosia artemisiifolia L. Atriplex patula L. Chenpodium album L. Digitaria sanguinalis (L.) Scop. Echinochloa crus-galli (L.) Beauv. Erigeron annuus (L.) Pers Conyza canadensis (L.) Cronq. Hibiscus trionum L. Panicum capillare L. Panicum dichotomiflorum Michx. Polygonum convolvulus L. Polygonum lapathifolium L. Polygonum persicaria L. Portulaca oleracea L. Setaria faberi Herrm. Setaria pumila (Poir.) Roemer & J. A. Schultes Setaria viridis (L.) Beauv. Sinapis arvensis L. Solanum ptychanthum Dun. Sonchus oleraceus L. Stellaria media (L.) Vill. Zea mays L.
Common name
Taxonomy Recruitment
Time to Predicted to anthesis Anemochory increase
Velvetleaf Pigweed, redroot Pigweed species Ragweed, common Spreading atriplex Lambsquarters, common Crabgrass, large Bamyardgrass Annual fleabane Canada fleabane Mallow, Venice Witchgrass Panicum, fall Buckwheat, wild Smartweed, pale Ladysthumb Purslane, common Foxtail, giant Foxtail, yellow Foxtail, green Mustard, wild Nightshade, eastern black Sowthistle, annual Chickweed, common Volunteer corn
Broadleaf Broadleaf Broadleaf Broadleaf Broadleaf Broadleaf Grass Grass Broadleaf Broadleaf Broadleaf Grass Grass Broadleaf Broadleaf Broadleaf Broadleaf Grass Grass Grass Broadleaf Broadleaf Broadleaf Broadleaf Grass
Medium Late Late Early Early Early Late Late Early Early Late Late Late Early Medium Medium Late Late Medium Late Early Medium Medium Early Late
Long Short Short Long Long Long Long Long Long Long Long Long Long Long Long Short Short Long Long Long Short Long Long Short Long
No No No No No No No No Yes Yes No No No No No No No No No No No No Yes No No
No No No No No No No No Yes Yes No Yes No No No No No No No No No No Yes No No
Annual, biennial, perennial DAUCA Daucus carota L. LACSE Lactuca serriola L. MATIN Tripleurospermum perforata (Merat) M. Lainz OXAST Oxalis stricta L. TRFSS Trifolium species
Carrot, wild Prickly lettuce Scentless chamomile
Broadleaf Broadleaf Broadleaf
Early Early Early
Long Long Long
No Yes No
No No No
Woodsorrel, yellow Clovers species
Broadleaf Broadleaf
Late Medium
Short Short
No No
No No
Perennial ASCSY CONAR CYPES PLAMA POAPR SONAR TAROF
Milkweed, common Bindweed, field Nutsedge, yellow Broad-leaved plantain Kentucky bluegrass Sowthistle, perennial Dandelion
Broadleaf Broadleaf Grass Broadleaf Grass Broadleaf Broadleaf
Late Medium Late Early Early Medium Late
Long Short Short Long Long Long Short
Yes No No No No Yes Yes
Yes No No No No No Yes
Asclepias syriaca L. Convolvulus arvensis L. Cyperus esculentus L. Plantago major L. Poa pratensis L. Sonchus arvensis L. Taraxacum officinale Weber ex Wiggers
system were more significant filters of the midseason weed community than rotation (Figure 1). The separation of glyphosate compared to conventional herbicide treatments in corn and soybean was more pronounced than in a previous study in GR canola conducted in western Canada (Harker et al. 2005). Distinct differences in weed communities between management with conventional herbicides and glyphosate observed in our study were strongly influenced by the shift from selective pre- to nonselective post-emergence applied herbicides which, for example, was not the case in Harker et al. (2005) in which conventional herbicides were also postemergence applied. However, differences between these studies may also have been influenced by the biome-specific weed species complement or the greater species frequency exclusion limit (25%) chosen by Harker et al. (2005). Curran et al. (2002) also suggested weed species shifts in glyphosateresistant corn and soybean. In the United Kingdom, Heard et al. (2006) reported that effects of herbicide-resistant cropping systems on weeds and invertebrates manifest quickly without further dramatic changes in the second year. Results from our study (data not shown) tend to support these observations. Redundancy analysis effectively separated the treatments (herbicide system, crop and rotation) and the first two
canonical variables were significant (P , 0.05) in all cases. The first two canonical variables explained a combined 72.5 to 89.1% of the total variation in the weed species observed among the treatments. The limited effect of rotation on midseason aboveground weed communities may have been influenced by the short duration of the experiment. In three experiments (Ridgetown NT, Woodstock NT, and Woodslee CT), treatment of corn and soybean with glyphosate resulted in more similar in-crop weed communities than when corn and soybean were treated with their respective conventional herbicides (Figure 1). Herbicide is an important filter for weed community assembly (Doucet et al. 1999) and, therefore, repeated application of the same selection pressure as in continuous GR corn and GR soybean was expected to lead to a more similar midseason weed community assembly in these treatments. At Ridgetown NT and Woodstock (NT and CT), midseason weed communities in corn and soybean treated with conventional herbicides were unique and clearly different. Swanton et al. (2006) reported similar results when examining the effects of various filters (combinations of tillage and herbicides in corn, soybean, and wheat) on midseason weed community assembly. Our experiments were not designed to test the effects of tillage system explicitly and no Gulden et al.: Glyphosate-resistant crops: weed communities
N
281
Figure 1. Biplots of the first two canonical variables of the species and treatment means at each location. For corn and soybean crops, main treatment codes indicate herbicide system (C 5 conventional herbicides, GR 5 glyphosate resistant), while subscript codes indicate the crop sequence beginning with the crop in which the measurement was taken (e.g., CWS 5 weed community in corn, preceded by wheat, preceded by soybean). The percentage of the data explained by each variable is indicated.
282
N
Weed Science 58, July–September 2010
Figure 2. Reaction norm plots showing the impact of herbicide system (Conv 5 conventional herbicides, GR 5 glyphosate resistant) on relative trait-densities in corn at five locations. Solid lines indicate statistically significant differences between herbicide systems, while dashed lines indicate no significant differences between herbicide systems. Traits investigated included: time of relative initiation of recruitment (recruitment early, medium, late), life cycle (annuals, perennials), taxonomic group (grasses), time to anthesis (time to anthesis short), anemochory (wind dispersal), and species thought to increase in GR systems including dandelion.
clear trends in weed community assembly among the treatments based on tillage system were observed. Tillage system interacts with other factors affecting weed community assembly (e.g., Mas and Verdu 2003). Similarities in species behavior in response to herbicide system were difficult to detect among the locations (Figure 1). The trait-based approach was used to facilitate comparison of herbicide system–induced weed trait shifts among locations with divergent weed populations that multivariate analysis was unable to detect. Impact of Herbicide System on the Frequency of TraitDensities of Weeds. Reduction in the complexity of the weed communities based on plant traits was a successful approach for testing hypotheses based on expected, herbicide-induced changes in weed community assembly based on trait-densities among divergent locations (Figures 2 and 3). Due to the lack of sufficient quantitative data for all weed species in this study, a qualitative approach to trait categorization was employed. Insufficient data for quantitative trait-based analysis is a recurring limitation (e.g., Hawes et al. 2005). Continuous, quantitative metrics of functional diversity in plants have been developed (Voille et al. 2007) and could be a powerful tool for weed science. Quantitative trait values, however, can be of equal magnitude within and among species and may respond to selection filters such as the timing of herbicide application (Hawes et al. 2005). The qualitative approach used here did
not account for within-species variation for trait-density determination nor potential treatment effects on trait values (selection or morphological plasticity). In our study, only traits with no or limited plasticity in response to treatments were examined. As predicted by Shaner (2000), treatment with glyphosate increased the proportion of late emerging weeds in the community in four of five experiments in both crops (Figures 2 and 3). Herbicide system, however, had little impact on the frequency of early recruiting weed species in the community. This result was not expected. The increases in the proportion of late emerging weeds which has been reported before (Hilgenfeld et al. 2004a,b; Puricelli and Tuesca 2005; Westra and Nissen, 2004) were almost exclusively due to a consistent decrease in the frequency of species with medium time of initiation of seedling recruitment. This was observed in soybean in all five and corn in three soybean experiments (Figures 2 and 3) and may be related to the timing of preemergence applications of glyphosate in GR systems. In our study, the early, medium, and late categories contained 17, 11, and 13 species, respectively. We are confident in accurate classification of weeds in the initiation of recruitment categories because values for most species were obtained from a direct comparison of these species under undisturbed conditions (Doll 2007). In our experiments, dandelion (Taraxacum officinale Weber ex Wiggers) was considered a species with late initiation of seedling recruitment. The time Gulden et al.: Glyphosate-resistant crops: weed communities
N
283
Figure 3. Reaction norm plots showing the impact of herbicide system (Conv 5 conventional herbicides, GR 5 glyphosate resistant) on relative trait-densities in soybean at five locations. Solid lines indicate statistically significant differences between herbicide systems, while dashed lines indicate no significant differences between herbicide systems. Traits investigated included: time of relative initiation of recruitment (recruitment early, medium, late), life cycle (annuals, perennials), taxonomic group (grasses), time to anthesis (time to anthesis short), anemochory (wind dispersal), and species thought to increase in GR systems including dandelion.
of seedling recruitment initiation in dandelion is different for overwintering rootstocks (early) and the new seed cohort (late) (Froese at al. 2005; Hacault and Van Acker 2006). In our experiments, midseason dandelion rosettes were small and almost exclusively originated from seed. An increase in the frequency of perennial weeds in GR compared to conventional herbicide systems was predicted (Culpepper 2006). This hypothesis could only be tested in four experiments due to an absence of life cycle diversity at Woodslee. An increase in perennials and a decrease in annual weed species was observed in three experiments in corn and two experiments in soybean (Figures 2 and 3), indicating that this prediction also held true in many instances. The magnitude of the shift in life histories was much more pronounced in corn compared to soybean indicating that weed spectrum, crop, and/or conventional herbicide system also play significant roles in filtering weed life cycles. Biennial species consistently contributed little (. 5% in most experiments) to the total weed community and were not affected substantially by herbicide system (data not shown). The effect of herbicide system on the prevalence of taxonomic groups in the midseason weed community seemed to be unique to tillage system. In NT experiments, a clear reduction in the frequency of grasses was observed in both crops when treated with glyphosate (Figures 2 and 3). In CT experiments, no changes in the frequency of grasses or 284
N
Weed Science 58, July–September 2010
broadleaved weeds were observed, with the exception of soybean at Woodstock where the frequency of grasses increased. In NT experiments, midseason weed densities of broadleaved weeds tended to be greater in the glyphosate than the conventional herbicide systems (Gulden et al. 2009) which suggests that the shift in taxonomic groups resulted from reduced efficacy of glyphosate on broadleaved species relative to grassy weeds. In GR cotton (Gossypium hirsutum L.), the lack of residual herbicides resulted in increases in the populations of annual grasses (Culpepper 2006). Herbicide system also induced shifts in the trait-densities of species with a short time to anthesis. Frequency of traitdensities of species with a time interval of 6 wk or less between recruitment and anthesis increased in both crops in NT experiments and in corn in two CT experiments. As expected, short time to anthesis was most closely associated with medium time of recruitment at Ridgetown NT and Woodstock NT and CT and with late time of emergence at the remaining locations (Figure 4). High efficacy of late in-crop applications of glyphosate would select for weed species with reduced time to complete their life cycle before harvest of the crop. In general, the relationships among traits at Ridgetown and the Woodstock locations were similar and differed substantially from those at Huron Park and Woodslee (Figure 4). This appeared to be influenced more by community structure than tillage system (see Gulden et al.
Figure 4. Dendrograms of relationships among traits for the total weed community at each location. Nine traits were compared including: early initiation of recruitment (Rec_E), medium initiation of recruitment (Rec_M), late initiation of recruitment (Rec_L), annual life cycle (Ann), biennial life cycle (Bien), perennial life cycle (Peren), grasses (Grass), short time to initiation of anthesis (TF_short), and anemochory (Disp_Win).
2009) and can lead to differential trait-density responses among locations (e.g., medium time to recruitment vs. perennial trait-density response at Huron Park compared to Ridgetown and Woodstock (Figures 2 and 3). The frequency of wind-dispersed species was predicted to increase in GR cropping systems (Shaner 2000). Indeed, this was observed in three of four experiments in corn and two of
four experiments in soybean with a substantially greater proportional change in trait-densities in corn (Figures 2 and 3). At Woodslee, no wind-dispersed weed species were present. Wind-dispersed species were closely related to perennial species at most locations (Figure 4). In our experiments, in-crop glyphosate applications selected for similar taxonomic groups and life cycles as reduced tillage Gulden et al.: Glyphosate-resistant crops: weed communities
N
285
Table 3. Bayer code, scientific name, common name, and frequency of occurrence in experimental units of weeds in each experiment in 2003 to 2005. Bayer code
Scientific name
Ridgetown NT
Common name
Woodstock NT
Woodstock Huron Park CT CT
Woodslee CT
Frequency Annual ABUTH AMARE AMASS AMBEL ATXPA CHEAL DIGSA ECHCG ERIAN ERICA HIBTR PANCA PANDI POLCO POLLA POLPE POROL SETFA SETLU SETVI SINAR SOLPT SONOL STEME ZEAMX
Abutilon theophrasti Medik. Amaranthus retroflexus L. Amaranthus species Ambrosia artemisiifolia L. Atriplex patula L. Chenpodium album L. Digitaria sanguinalis (L.) Scop. Echinochloa crus-galli (L.) Beauv. Erigeron annuus (L.) Pers. Conyza canadensis (L.) Cronq. Hibiscus trionum L. Panicum capillare L. Panicum dichotomiflorum Michx. Polygonum convolvulus L. Polygonum lapathifolium L. Polygonum persicaria L. Portulaca oleracea L. Setaria faberi Herrm. Setaria pumila (Poir.) Roemer & J. A. Schultes Setaria viridis (L.) Beauv. Sinapis arvensis L. Solanum ptychanthum Dun. Sonchus oleraceus L. Stellaria media (L.) Vill. Zea mays L.
Velvetleaf Pigweed, redroot Pigweed species Ragweed, common Spreading atriplex Lambsquarters, common Crabgrass, large Barnyardgrass Annual fleabane Canada fleabane Mallow, Venice Witchgrass Panicum, fall Buckwheat, wild Smartweed, pale Ladysthumb Purslane, common Foxtail, giant Foxtail, yellow Foxtail, green Mustard, wild Nightshade, eastern black Sowthistle, annual Chickweed, common Volunteer corn
Annual, biennial, perennial DAUCA LACSE MATIN OXAST TRFSS
Daucus carota L. Lactuca serriola L. Matricaria indora L. Oxalis stricta L. Trifolium species
Carrot, wild Prickly lettuce Scentless chamomile Woodsorrel, yellow Clover species
Perennial ASCSY CONAR CYPES PLAMA POAPR SONAR TAROF
Asclepias syriaca L. Convolvulus arvensis L. Cyperus esculentus L. Plantago major L. Poa pratensis L. Sonchus arvensis L. Taraxacum officinale Weber ex Wiggers
Milkweed, common Bindweed, field Nutsedge, yellow Broad-leaved plantain Kentucky bluegrass Sowthistle, perennial Dandelion
systems (Derksen et al. 1993; Froud-Williams et al. 1981; Thomas et al. 2004). Using the trait-based approach, the response in the cumulative frequency of the group of species suggested in the literature to increase in the glyphosate treatment was tested (Table 2) (Duncan and Weller 1987; Hacault and Van Acker 2006; King et al. 2004; Krausz et al. 1996; Marshall et al. 2000; Nandula et al. 2005; Owen and Zelaya 2005; Patzoldt et al. 2002; Shaner 2000; VanGessel 2001). The effect of glyphosate on this group of species, however, was small compared to the response in other traits (Figures 2 and 3). Overall, an increase in the cumulative frequency of these species was observed only 40% of the time. Examination of this lack of a cumulative response to glyphosate at the species level shows a clear displacement of species within this group resulting in no net increase in the cumulative frequency. For example, in corn at Ridgetown NT, velvetleaf (Abutilon theophrasti Medik.) was replaced by dandelion in the glyphosate treatments (Figure 1) with both species contributing more than 50% to the total weed densities in these respective treatments (data not shown). Similarly, a large 286
N
Weed Science 58, July–September 2010
0.55 0.48 0.48 0.11 0.83 0.29
0.08 0.64 0.05
0.58 0.05
0.50 0.29 0.19 0.04
0.72 0.29 0.33 0.10
0.23
0.37
0.50
0.73 0.34
0.73
0.25
0.64
0.65
0.18 0.44 0.28 0.39 0.17 0.17 0.05 0.28 0.47 0.14 0.31 0.13 0.06
0.27
0.53 0.14
0.06
0.26 0.13
0.69
0.11 0.33
0.23
0.16 0.41
0.07
0.08
0.26 0.03 0.08
0.77 0.14 0.13
0.06 0.02
0.40
0.38 0.14
0.05 0.58
0.11 0.13
0.03
0.20 0.10
0.96
0.99
0.67
0.08 0.14 0.06
proportion of the triazine-resistant population of common lambsquarters was replaced by dandelion in the glyphosate treatment in corn at Woodstock in both tillage systems (Figure 1). Multivariate analysis revealed similar results with few species expected to increase in GR systems showing preferential association with this system (Figure 1). Canada fleabane, for example, was present at three locations and was associated most strongly with winter wheat in the GR system at Ridgetown NT (data not shown). In the Woodstock experiments, Canada fleabane was found sporadically in corn and winter wheat (data not shown) in the GR herbicide system only. Another example species is common lambsquarters which recently was observed to increase in GR cropping systems (Westra et al. 2008). A similar consistent response of this species was not observed in our experiments. While lambsquaters was associated with the GR cropping systems Huron Park CT and Woodslee CT, this species associated most closely with corn and soybean treated with conventional herbicides at Ridgetown NT (Figure 1). At Woodstock, the population of lambsquaters was resistant to triazines and as a
result was most closely associated with corn treated with conventional herbicides. The results clearly show that trait-based reduction of weed community complexity allowed for a direct comparison of weed communities with vastly different quantitative and qualitative characteristics. A total of 37 weed species were recorded over the 3-yr period on the five experiments evaluated, with diverse midseason weed communities ranging from 8 to 25 species (Table 3). At Woodslee CT, where weed species richness was lowest (Gulden et al. 2009), cluster analysis revealed lack of clear separation among traits (Figure 4). This contributed to directional similarities in the proportional change in trait-densities among correlated traits (Figures 2 and 3). For example, in corn treated with glyphosate at Ridgetown and Woodstock, the increase in frequency of perennials was associated with a frequency increase in anemochorous species and a reduction in frequency of wind-pollinated and grassy species. The high frequency of dandelion in the community contributed to these relationships. Although arguably crude, this nominal traitbased approach to investigating the weed community nevertheless allowed for the testing of hypotheses on the impacts of GR cropping systems. Despite limitations of a priori creation of artificial trait categories, this analysis revealed valuable information about the midseason weed community structure in these management systems. In conclusion, trait-based reduction of complexity in weed communities provided valuable information with respect to the impact of GR cropping systems on the midseason weed community. Trait-densities based on nominal categorization revealed general similarities among diverse weed communities and the significance of herbicide system as a filter for several traits; however, not all weed communities responded similarly to these selection pressures. Trait-density– based evaluation of the weed community was most appropriate for more complex weed communities, i.e., those not dominated by few species. In the future, trait-based approaches to weed community assembly may be useful for designing effective and efficient integrated weed management systems; however, more research is required to identify, quantify, and categorize relevant traits.
Sources of Materials 1
Statistical software, SAS Institute, 2002-2003, SAS Software, Version 9.1, SAS Institute, Cary, NC 57513.
Acknowledgments The authors wish to thank Monsanto Canada, Inc. for providing the funding and donating the seed for this research. The authors also like to thank OMAFRA (Ontario Ministry of Agriculture, Food, and Rural Affairs) as well as K. Chandler, C. Shropshire, T. Cowan, M. Whaley, and numerous summer students for their technical assistance.
Literature Cited Alex, J. F. and C. M. Switzer. 1975. Ontario Weeds. Publication 505. Ministry of Agriculture, Food and Rural Affairs, Ontario, Canada. Toronto, ON, Canada: Queen’s Printer for Ontario. 200 p.
Anonymous. 2004. Ontario field crops research and service committee annual report. http://www.ontla.on.ca/library/respository/ser/10316602/2004.pdf. Accessed: April 3, 2009. Baucom, R. S. and R. Mauricio. 2004. Fitness costs and benefits of novel herbicide tolerance in a noxious weed. Proc. Natl. Acad. Sci. 101: 13386–13390. Booth, B. D. and C. J. Swanton. 2002. Assembly theory applied to weed communities. Weed Sci. 50:2–13. Bryson, C. T. and G. D. Wills. 1985. Susceptibility of bermudagrass (Cynodon dactylon) biotypes to several herbicides. Weed Sci. 33:848–852. Cavers, P. B. 1995. The biology of Canadian weeds—Contributions 62–83. Ottawa, ON, Canada: The Agricultural Institute of Canada. 338 p. Cavers, P. B. 2000. The biology of Canadian weeds—Contributions 84–102. Ottawa, ON, Canada: The Agricultural Institute of Canada. 338 p. Cavers, P. B. 2005. The biology of Canadian weeds—Contributions 103–129. Ottawa, ON, Canada: The Agricultural Institute of Canada. 516 p. Choler, P. 2005. Consistent shifts in Alpine plant traits along a mesotopographical gradient. Arct. Antarct. Alp. Res. 37:444–453. Cingolani, A. M., G. Posse, and M. B. Collantes. 2005. Plant functional traits, herbivore selectivity and response to sheep grazing in Patagonian steppe grasslands. J. Appl. Ecol. 42:50–59. Culpepper, A. S. 2006. Glyphosate induced weed shifts. Weed Technol. 20:277–281. Curran, W. S., K. Handwerk, and D. D. Lingenfelter. 2002. Temporal weed dynamics as influenced by corn and soybean herbicides. Weed Sci. Soc. Am. Abstracts 42:6. De Bello, F., J. Leps, and M. T. Sebastia. 2005. Predictive value of plant traits to grazing along a climatic gradient in the Mediterranean. J. Appl. Ecol. 42:824–833. Derksen, D. A., G. P. Lafond, A. G. Thomas, H. A. Loeppky, and C. J. Swanton. 1993. Impact of agronomic practices on weed communities—tillage systems. Weed Sci. 41:409–417. Diaz, S., A. Acosta, and M. Cabido. 1992. Morphological analysis of herbaceous communities under different grazing regimes. J. Veg. Sci. 3:689–696. Diaz, S., S. Lavorel, S. McIntyre, V. Falczuk, F. Casanoves, D. G. Milchunas, C. Skarpe, G. Rusch, M. Sternberg, I. Noy-Meir, J. Landsberg, W. Zhang, H. Clark, and B. D. Campbell. 2007. Plant trait responses to grazing—a global synthesis. Glob. Change Biol. 13:313–341. Doll, J. 2007. Knowing when to look for what: Weed emergence and flowering sequences in Wisconsin. http://128.104.239.6/uw_weeds/extension/articles/ weedemerge.htm. Accessed: February 27, 2010. Doucet, C., S. E. Weaver, A. S. Hamill, and J. Zhang. 1999. Separating the effects of crop rotation from weed management on weed density and diversity. Weed Sci. 47:729–735. Duncan, C. N. and S. C. Weller. 1987. Heritability of glyphosate susceptibility among biotypes of field bindweed. J. Hered. 78:257–260. Fried, G., B. Chauvel, and X. Reboud. 2009. A functional analysis of large-scale temporal shifts from 1970 to 2000 in weed assemblages of sunflower crops in France. J. Veg. Sci. 20:49–58. Fried, G., L. R. Norton, and X. Reboud. 2008. Environmental and management factors determining weed species composition and diversity in France. Agric. Ecosyst. Environ. 128:68–76. Froese, N. T., R. C. Van Acker, and L. F. Friesen. 2005. Influence of spring tillage and glyphosate treatment on dandelion (Taraxacum officinale) control in glyphosate-resistant canola. Weed Technol. 19:283–292. Froud-Williams, R. J., R. J. Chancellor, and D.S.H. Drennan. 1981. Potential changes in weed floras associated with reduced cultivation systems for cereal production in temperate regions. Weed Res. 21:99–109. Gulden, R. H., P. H. Sikkema, A. S. Hamill, F. J. Tardif, and C. J. Swanton. 2009. Glyphosate-resistant cropping systems in Ontario: Weed control, diversity, and yield. Weed Sci. 57:665–672. Hacault, K. M. and R. C. Van Acker. 2006. Emergence timing and control of dandelion (Taraxacum officinale) in spring wheat. Weed Sci. 54:172–181. Harker, K. N., G. W. Clayton, R. E. Blackshaw, J. T. O’Donovan, N. Z. Lupwayi, E. N. Johnson, G. P. Lafond, and R. B. Irvine. 2005. Glyphosateresistant spring wheat production system effects on weed communities. Weed Sci. 53:451–464. Hawes, C., G. S. Begg, G. R. Squire, and P.P.M. Iannetta. 2005. Individuals as the basic accounting unit in studies of ecosystem function: functional diversity in shepherd’s purse, Capsella. Oikos 109:521–534. Heard, M. S., S. J. Clark, P. Rothery, J. N. Perry, D. A. Bohan, D. R. Brooks, G. T. Champion, A. M. Dewar, G. Hawes, A. J. Haughton, M. J. May, R. J. Scott, R. S. Stuart, G. R. Squire, and L. G. Firbank. 2006. Effects of successive seasons of genetically modified herbicide-tolerant corn cropping on weeds and invertebrates. Ann. Appl. Biol. 149:249–254. Heard, M. S., C. Hawes, G. T. Champion, S. J. Clark, L. G. Firbank, A. J. Haughton, A. M. Parish, J. N. Perry, P. Rothery, D. B. Roy, R. J. Scott, M. P.
Gulden et al.: Glyphosate-resistant crops: weed communities
N
287
Skellern, G. R. Squire, and M. O. Hill. 2003. Weeds in fields with contrasting conventional and genetically modified herbicide-tolerant crops. II. Effects on individual species. Philos. Trans. R. Soc. Lond. B Biol. Sci. 358:1833–1846. Hilgenfeld, K. L., A. R. Martin, D. A. Mortensen, and S. C. Mason. 2004a. Weed management in glyphosate resistant soybean system: Weed emergence patterns in relation to glyphosate treatment timing. Weed Technol. 18:277–283. Hilgenfeld, K. L., A. R. Martin, D. A. Mortensen, and S. C. Mason. 2004b. Weed management in glyphosate resistant soybean system: weed species shifts. Weed Technol. 18:284–291. King, S. R., E. S. Hagwood, and J. H. Westwood. 2004. Differential response of a common lambsquarters (Chenopodium album) biotype to glyphosate. Weed Sci. Soc. Am. Abstracts 44:68. Krausz, R. G., G. Kapusta, and J. L. Matthews. 1996. Control of annual weeds with glyphosate. Weed Technol. 10:957–962. Leps, J. and P. Smilauer. 1999. Multivariate analysis of ecological data University of South Bohemia, Ceske Budejovice, Czech Republic. 110 p. Lindborg, R. and O. Eriksson. 2005. Functional response to land use change in grasslands: Comparing species and trait data. Ecosci. 12:183–191. Louault, F., V. D. Pillar, J. Aufrere, E. Garnier, and J. F. Soussana. 2005. Plant traits and functional types in response to reduced disturbance in a semi-natural grassland. J. Veg. Sci. 16:151–160. Marshall, M. W., K. Al-Khatib, and L. Maddux. 2000. Weed community shifts associated with continuous glyphosate applications in corn and soybean rotation. Pages 22–25 in Proceedings of the Western Society of Weed Science 53. Las Cruces, NM: Western Society of Weed Science. Mas, M. T. and A.M.C. Verdu. 2003. Tillage system effects on weed communities in a 4-year crop rotation under Mediterranean dryland conditions. Soil Till. Res. 74:15–24. Mayo, C. M., M. J. Horak, D. E. Peterson, and J. F. Boyer. 1995. Differential control of four Amaranthus species by six postemergence herbicides in soybean (Glycine max). Weed Technol. 9:141–147. Mulligan, G. A. 1979. The biology of Canadian weeds—Contributions 1–32. Publication 1693. Ottawa, ON, Canada: Communications Branch, Agriculture and AgriFood Canada. 380 p. Mulligan, G. A. 1984. The biology of Canadian weeds—Contributions 33–61. Publication 1765. Ottawa, ON, Canada: Communications Branch, Agriculture and AgriFood Canada. 415 p. Nandula, V. K., K. N. Reddy, S. O. Duke, and D. H. Poston. 2005. Glyphosateresistant weeds: current status and future outlook. Out. Pest Manage. 16:183–187. Økland, R. H. 2003. Partitioning the variation in a plot-by-species data matrix that is related to n sets of explanatory variables. J. Veg. Sci. 14:693–700. Owen, M.D.K. and I. A. Zelaya. 2005. Herbicide-resistant crops and weed resistance to herbicides. Pest Manage. Sci. 61:301–311. Patzoldt, W. L., P. J. Tranel, and A. G. Hager. 2002. Variable herbicide response among Illinois waterhemp (Amaranthus rudis and A. tuberculatus) populations. Crop Prot. 21:707–712.
288
N
Weed Science 58, July–September 2010
Puricelli, E. and D. Tuesca. 2005. Weed density and diversity under glyphosateresistant crop sequences. Crop Prot. 24:533–542. Pykala, J. 2004. Cattle grazing increases plant species richness of most species trait groups in mesic semi-natural grasslands. Plant Ecol. 175:217–226. Shaner, D. L. 2000. The impact of glyphosate-tolerant crops on the use of other herbicides and on resistance management. Pest Manage. Sci. 56: 320–326. Sikkema, P. H. and N. Soltani. 2007. Herbicide-resistant crops in eastern Canada. In R. H. Gulden and C. J. Swanton, eds. The First Decade of Herbicide Resistant Crops in Canada. Topics in Canadian Weed Science, Volume 4. 3–13. Saint-Anne-de Bellevue, Que´bec, Cananda: Canadian Weed Science Society–Socie´te´ canadienne de malherbologie. Storkey, J. 2006. A functional group approach to the management of UK arable weeds to support biological diversity. Weed Res. 46:513–522. Swanton, C. J., B. D. Booth, K. Chandler, D. R. Clements, and A. L. Shrestha. 2006. Management in a modified no-tillage corn-soybean-wheat rotation influences weed population and community dynamics. Weed Sci. 54: 47–58. Swanton, C. J., A. L. Shrestha, R. C. Roy, B. R. Ball-Coelho, and S. Z. Knezevic. 1999. Effect of tillage systems, N, and cover crop on the composition of weed flora. Weed Sci. 47:454–461. Sweat, J. K., M. J. Horak, D. E. Peterson, R. W. Lloyd, and J. E. Boyer. 1998. Herbicide efficacy on four Amaranthus species in soybean (Glycine max). Weed Technol. 12:315–321. Thomas, A. G., D. A. Derksen, R. E. Blackshaw, R. C. Van Acker, A. Legere, P. R. Watson, and G. C. Turnbull. 2004. A multistudy approach to understanding weed population shifts in medium- to long-term tillage systems. Weed Sci. 52:874–880. VanGessel, M. J. 2001. Glyphosate-resistant horseweed from Delaware. Weed Sci. 49:703–705. Voille, C., M. L. Naves, D. Ville, E. Kazakou, C. Fortunel, I. Hummel, and E. Garnier. 2007. Let the concept of trait be functional! Oikos 16:882–892. Weiher, E., A. van der Werf, K. Thompson, M. Roderick, E. Garnier, and O. Eriksson. 1999. Challenging Theophrastus: a common core list of plant traits for functional ecology. J. Veg. Sci. 10:609–620. Westra, P. and S. Nissen. 2004. Weed population dynamics in conventional and Roundup Ready irrigated crops. Weed Sci. Soc. Am. Abstracts 44:125. Westra, P., R. Wilson, S. D. Miller, P. W. Stahlman, G. W. Wicks, P. L. Chapman, J. Withrow, D. Legg, C. Alford, and T. A. Gaines. 2008. Weed population dynamics after six years under glyphosate- and conventional herbicide-based weed control strategies. Crop Sci. 48:1170–1177. Wiegmann, S. M. and D. M. Waller. 2006. Fifty years of change in northern upland forest understories: Identity and traits of ‘‘winner’’ and ‘‘loser’’ plant species. Biol. Conserv. 129:109–123.
Received December 21, 2009, and approved April 7, 2010.