Introduction
Snap beans are cultivars of common bean (Phaseolus vulgaris L.) grown for their young and unripe fruits (pods). More than 80% of snap bean is grown commercially for processing, with the remainder grown for the fresh market (Davis et al. Reference Davis, Weber, Wechsler, Lucier, Wakefield and Vaiknoras2023). Two-thirds of processed snap bean is canned and one-third is frozen (USDA-NASS 2024). Snap bean in the United States is grown for processing in the Northeast (NE), the Midwest (MW), and the Northwest (NW) (USDA-NASS 2024). In the last decade, snap bean grown for processing has decreased ∼30% (USDA-NASS 2024). Declining production is attributed to increased snap bean imports and changing consumer preference toward fresh and frozen products (Davis et al. Reference Davis, Weber, Wechsler, Lucier, Wakefield and Vaiknoras2023).
Among the biggest challenges in row crop production in North America are weeds, whose competition for resources, costs of control, and harvest interference equate to an estimated US$44 billion in economic losses annually (Soltani et al. Reference Soltani, Dille, Burke, Everman, VanGessel, Davis and Sikkema2016, Reference Soltani, Dille, Burke, Everman, VanGessel, Davis and Sikkema2017). In snap bean production, competition from weeds escaping control causes up to 80% in direct yield loss (Odero and Wright Reference Odero and Wright2018; Qasem Reference Qasem1995). Certain species cause greater harm by contaminating the harvested product. For instance, berries of nightshade species (Solanum spp.) can lead to entire harvested loads being rejected by processors (Peachey Reference Peachey2019). Consumers, vegetable processors, and food inspectors alike have a low tolerance for weedy vegetation (i.e., foreign material) in snap bean products. Despite the significance of weeds in snap bean production, detailed knowledge of specific weed problems in snap bean is limited to anecdotal observations. Weed species or species-groups observed in snap bean trials and other reports in the 21st century include amaranth species (Amaranthus spp.), common ragweed (Ambrosia artemisiifolia L.), common lambsquarters (Chenopodium album L.), nutsedge species (Cyperus spp.), large crabgrass [Digitaria sanguinalis (L.) Scop.], morningglory species (Ipomoea spp.), common purslane (Portulaca oleracea L.), wild radish (Raphanus raphanistrum L.), and hairy nightshade (Solanum physalifolium Rusby) (Aguyoh et al. Reference Aguyoh and Masiunas2003a, Reference Aguyoh and Masiunas2003b; Bailey et al. Reference Bailey, Wilson and Hines2003; Boyhan et al. Reference Boyhan, Harrison, Sumner, Langston, Sparks, Culpepper, Hurst, MacLean, Li, Hawkins and Fonsah2013; Bradley et al. Reference Bradley, Bellinder and Brainard2007; Peachey Reference Peachey2019; Peachey et al. Reference Peachey, William and Mallory-Smith2004; Van Wychen Reference Van Wychen2022).
Quantitative knowledge of weed community structure is fundamental to understanding the scope of weed issues and directing future management and research efforts in numerous row crops (Frick and Thomas Reference Frick and Thomas1992; Froud-Williams and Chancellor Reference Froud-Williams and Chancellor1982; Rydberg and Milberg Reference Rydberg and Milberg2000; Salonen et al. Reference Salonen, Hyvönen and Jalli2001; Thomas Reference Thomas1985; Thomas and Dale Reference Thomas and Dale1991; Williams et al. Reference Williams, Rabaey and Boerboom2008). Weeds observed late in the crop growing season, hereafter called residual weeds, are the cumulative result of unsuccessful weed management. Differences in residual weeds occurring across growing regions reflect the influence of variable environmental conditions and management practices. These residual populations also contribute to the weed seedbanks of their respective fields. Collectively, the canopy of the weed community at harvest represents the greatest threat to contaminating a snap bean product with foreign material. The objectives of this work were to (1) quantify the residual weed community in snap bean grown for processing across the major growing regions in the United States and (2) investigate linkages between the density of residual weeds and their contributions to the weed canopy cover.
Materials and Methods
Description of Survey Area and Survey Methodology
Surveys of weeds in snap bean fields were conducted in 2019 to 2023 across the NW, MW, and NE regions. The authors collaborated with vegetable processors in each region to identify candidate fields. Fields were selected from counties that were among the leading snap bean producers for each state (USDA-NASS 2024; Figure 1). Furthermore, surveys were conducted across a broad period of snap bean harvest, from June through October.
Figure 1. Counties of surveyed snap bean fields in the (A) Northwest, (B) Midwest, and (C) Northeast regions of the United States.
Surveys were conducted within 1 wk before harvest. The survey methodology utilized the approach described by Thomas (Reference Thomas1985) with slight modifications. In each field, 30 quadrats, or 1 quadrat ha−1 on fields ≤20 ha, were placed randomly along a 300- to 500-m loop across the field. The minimum quadrat size was 0.5 m2 (1-m length and 0.5-m width), and quadrats were placed parallel to the crop rows. Field areas within 20 m of the field edge were avoided. In each quadrat, residual weeds were enumerated by species. Weed species belonging to the same genus were similar in appearance in the seedling stage.
Quantifying the contribution of each weed species or species-group to the overall weed canopy at the time of crop harvest was not practical. Therefore, a visual estimate of the total cover of the weed community canopy, expressed as a percentage of the quadrat, was recorded in each quadrat.
Data Analysis
Quantitative indices of field frequency, mean field density, mean occurrence field density, mean field uniformity, and mean occurrence field uniformity were calculated (Thomas Reference Thomas1985). Field frequency (F) is the number of fields in which a species or species-group occurred, expressed as a percentage of the total number of fields. Mean field density (MFD) is the average number of individuals of a species or species-group k m−2 across all fields. Mean occurrence field density (MOFD) is the average number of individuals of a species or species-group k m−2 in fields where the species or species-group k occurred. Mean field uniformity (MFU) is the average number of quadrats in which a species or species-group k occurred across all fields, expressed as a percentage of all quadrats. Mean occurrence field uniformity (MOFU) is the average number of quadrats in which a species or species-group k occurred, expressed as a percentage of quadrats in fields where species or species-group k occurred. Total weed cover of all quadrats was averaged for each field to obtain the mean weed cover (MWC).
To rank the contribution of a species or species-group, F, MFD, and MFU were combined into a single index called relative abundance (RA) (Thomas Reference Thomas1985). RA assumes F, MFD, and MFU all have equal contribution to the weed community and has no units. Every weed species or species-group k value of F, MFD, and MFU was divided by its respective sum values of F, MFD, and MFU of all species or species-groups. The obtained values of each species or species-group k represent the relative values of F, MFD, and MFU, and the relative values of all species or species-groups is 100. Therefore, as there are three indices, the total value of RA of all species or species-groups is 300.
The Simpson index was used to characterize the diversity of the weed community (Simpson Reference Simpson1949). The Simpson index is defined as the probability that two individuals chosen at random and independently from an infinitely large population will belong to the same group. In this instance, the group represents a weed species or species-group. The index value is expressed as a reciprocal called Simpson’s reciprocal index (SRI). Higher values of the index represent more diversity, with the lowest value being 1 (representing a community dominated by only one species) and the highest value being the total number of species in the community, which are all evenly distributed. Hence, the SRI is seen as a dominance index by giving more weight to the dominant species of the community (Kent Reference Kent2012).
Distributions of the data did not meet the assumptions of normality and homoscedasticity necessary to conduct the parametric t-test; therefore, the nonparametric Wilcoxon-Mann-Whitney U-test was utilized (Fay and Proschan Reference Fay and Proschan2010; Mann and Whitney Reference Mann and Whitney1947) to compare regions for SRI, MFD, MFC, and MFU.
A random forest algorithm was used to gain insight into relationships between density of each species or species-group and weed cover (Breiman Reference Breiman2001). Random forest is a classifier consisting of a collection of tree-structured classifiers wherein each tree casts a unit vote for the most popular class of input. Random forest is nonparametric, so classical regression assumptions relating to data structure and distribution are not required. The goal of this approach was to determine the weed species or species-group that best predicted weed cover. Species or species-groups contributing to weed cover provide additional information on the significance of a weed species or species-group in the weed community (Kent Reference Kent2012; Nkoa et al. Reference Nkoa, Owen and Swanton2015). This is particularly important for a crop like snap bean, for which weedy vegetation at crop harvest can contaminate the food product.
The package ranger in R statistical software (v. 4.3.2; R Core Team 2023) was used for the random forest analysis. Tuning parameters were set so that the model with the lowest root mean-square error and highest goodness of fit (pseudo-R2) values could be fit. The number of individual regression trees was set to 1,000, as suggested by Breiman (Reference Breiman2001), while the optimal number of independent variables randomly selected as candidates for each split in the trees was set to 55, and the minimum optimal number of observations in each terminal node was set to 5. Across all regions, 9,999 quadrat samples were collected; however, due to missing data, the actual sample size was 8,178 quadrats. The dataset used for training the model was 80% of the whole sample size (6,542 quadrats), while the remaining 20% (1,636 quadrats) was used as a test set for checking the accuracy of the trained model. Before model fitting, the values of all variables were both scaled and transformed with the Yeo-Johnson transformation (Yeo and Johnson Reference Yeo and Johnson2000). Due to certain randomization aspects of the algorithm, it is only possible to determine how important certain predictors were in the model, not necessarily what kind of relationship they have with the response variable. The “importance” of the predictor variable is defined as permutation importance, which considers the positive effect it had on the prediction performance (Breiman Reference Breiman2001). Partial dependence plots were used to visualize model relationships.
Results and Discussion
Overview
A total of 358 snap bean fields were surveyed. The majority (57%) were from the MW region, followed by the NE (23%) and the NW (20%). Fields from these three regions reflect the wide range of environmental conditions where snap bean is grown for processing in the United States. The MW and NE mostly have a humid continental climate (Köppen climate types Dfa and Dfb) with temperatures that vary greatly from summer to winter and appreciable precipitation (Paleontological Research Institution 2022; Wall and Parrish Reference Wall and Parrish2014). In the NW, the vast majority of the fields were surveyed in the western part of Oregon state, which has a warm-summer Mediterranean climate (Köppen climate type Csb) characterized by warm and dry summers and mild to cool and wet winters (Zabel et al. Reference Zabel, Parrish, Moore and Lewis2014).
A total of 109 residual weed species or species-groups representing 31 plant families were observed. The NE had the most weed species or species-groups and families, 90 and 30, respectively. The MW had 58 weed species or species-groups from 24 families. The NW had 57 species or species-groups from 20 families. There were 26 shared weed species or species-groups among the three regions, 21 shared species or species-groups between NE and MW, and NE and NW, and only 2 shared species or species-groups between MW and NW (Figure 2). There were 22 species or species-groups that occurred only in the NE region, while there were 9 and 8 species or species-groups occuring only in the MW and NW regions, respectively. Even though the number of observed residual weed species or species-groups exceeds 100, in ecological surveys of plant communities, it is usual for only a few species to dominate individual sites (Kent Reference Kent2012), particularly in agroecosystems where intense selection pressure from management influences the weed community (Storkey and Neve Reference Storkey and Neve2018). This is evidenced in the diversity of weed communities, as characterized by field SRIs. While several species or species-groups were observed throughout each region, most individual fields had an SRI of 1 to 3 (Figure 3A), indicating that one to three species often dominated the residual weed community.
Figure 2. Venn diagram of the number of unique and shared weed species or species-groups among Northwest, Midwest, and Northeast regions of the United States.
Figure 3. Density distribution plots of (A) Simpsons’s reciprocal index, (B) mean field weed density, (C) mean field weed uniformity, and (D) mean field weed cover across regions.
A majority (95%) of fields had residual weeds at the time of snap bean harvest. Several quantitative indices used to characterize the weed community were skewed right. For instance, it was not uncommon to have MFD >10 plants m−2 and MWC > 5% (Figure 3B and 3C, respectively). The extent to which these residual weed communities resulted in yield losses or contaminated harvested product is unknown. However, incomplete weed control during crop flowering can exacerbate the effects of adverse weather on crop yield (Konsens et al. Reference Konsens, Ofir and Kigel1991; Landau et al. Reference Landau, Hager and Williams2021, Reference Landau, Hager and Williams2022).
There were some important specific distinctions in the residual weed community of each production region. Diversity was highest (P-value < 0.01) in the MW and NE (median SRI values of 1.46 and 1.40, respectively) compared with the NW (median SRI value of 1.22) (Figure 4A). This was due to several fields in the MW and NE having more than two dominant weed species or species-groups. The MW had a higher median MFD (10.8 plants m−2), MWC (6.6%), and MFU (86.7%) than the other two regions (P-value < 0.01) (Figure 4B–D). Collectively, the NW had among the smallest residual weed communities for every metric measured.
Figure 4. Distributions of (A) Simpson’s reciprocal index, (B) mean field weed density, (C) mean field weed cover, and (D) mean field weed uniformity by region. Values above each box-and-whisker plot represent median values with median absolute deviation in parentheses.
Weed Community Composition
Amaranth species and C. album were among the most abundant species or species-groups observed in the NW (Table 1), MW (Table 2), and NE (Table 3). Waterhemp [Amaranthus tuberculatus (Moq.) Sauer] is a threat to snap bean production, because the stems break into pod-sized fragments at crop harvest and contaminate harvested product (R Pequinot, personal communication). Aguyoh and Masiunas (Reference Aguyoh and Masiunas2003b) noted that redroot pigweed (Amaranthus retroflexus L.) was becoming more common in MW snap bean production and showed the weed caused >50% yield loss when crop and weed emergence coincided. Other amaranth species observed in the survey of growers’ fields included smooth pigweed (Amaranthus hybridus L.) and Palmer amaranth (Amaranthus palmeri S. Watson). Amaranth species will continue to be an issue in snap bean production due to their high level of adaptability, widespread herbicide resistance, enormous seed production, and presence throughout the growing season due to discontinuous germination. Chenopodium album has been identified as a troublesome weed in snap bean production in New York (Van Wychen Reference Van Wychen2022) and Oregon (Peachey Reference Peachey2019). Chenopodium album has been the object of study in snap bean dating to the mid-1980s (Vencill et al. Reference Vencill, Wilson and Hines1990; Wilson and Hines Reference Wilson and Hines1987). Wilson and Hines (Reference Wilson and Hines1987) evaluated the use of acifluorfen for postemergence control of C. album in snap bean production. Meanwhile, Vencill et al. (Reference Vencill, Wilson and Hines1990) evaluated the effectiveness of imazethapyr use in snap bean production. Both the works of Wilson and Hines (Reference Wilson and Hines1987) and Vencill et al. (Reference Vencill, Wilson and Hines1990) had the goal of finding new herbicide options for C. album control in snap bean production, since it was noted as a predominant weed in snap bean production in Virginia.
Table 1. Top 30 weed species in the Northwest region arranged by relative abundance a .
a Abbreviations: F, field frequency; MFD, mean field density; MFU, mean field uniformity; MOFD, mean occurrence field density; MOFU, mean occurrence field uniformity; RA, relative abundance.
Table 2. Top 30 weed species in the Midwest region arranged by relative abundance a .
a Abbreviations: F, field frequency; MFD, mean field density; MFU, mean field uniformity; MOFD, mean occurrence field density; MOFU, mean occurrence field uniformity; RA, relative abundance.
Table 3. Top 30 weed species in the Northeast region arranged by relative abundance a .
a Abbreviations: F, field frequency; MFD, mean field density; MFU, mean field uniformity; MOFD, mean occurrence field density; MOFU, mean occurrence field uniformity; RA, relative abundance.
Nightshade species such as black nightshade (Solanum nigrum L.), S. physalifolium, and eastern black nightshade (Solanum ptychanthum Dunal) were observed throughout the United States, ranking as high as fourth and third in relative abundance in the NW (Table 1) and MW (Table 2), respectively. Peachey (Reference Peachey2019) reported these species as being very troublesome in Oregon, as their berries, toxic to humans, can contaminate the harvested product, leading to entire harvested loads being rejected at processing facilities.
Across the United States, D. sanguinalis was one of the most abundant grass weeds in snap bean (Tables 1–3). In the early 2000s, D. sanguinalis was observed as a weed in MW snap bean production and could cause >50% yield loss at densities as low as 2 plants m−2 (Aguyoh and Masiunas Reference Aguyoh and Masiunas2003a). In the MW and NE regions, MOFD of D. sanguinalis exceeded >12 plants m−2, suggesting the weed may be widely troublesome.
Common chickweed [Stellaria media (L.) Vill.] was often observed in snap bean throughout the United States (Tables 1–3). Stellaria media was mainly observed in the seedling stage in fields surveyed late in the growing season (P.P., personal observation). Because S. media is a winter annual, the relatively high MOFD of its seedlings at crop harvest likely presents a minimal threat to snap bean production. Additional winter annuals that were often observed in the seedling stage in late-harvested snap bean fields of two or more regions included henbit (Lamium amplexicaule L.) and shepherd’s purse [Capsella bursa-pastoris (L.) Medik.].
Carpetweed (Mollugo verticillata L.) and ivyleaf morningglory (Ipomoea hederacea Jacq.), present in the MW and NE regions, were among the most abundant species in those two regions (Tables 2 and 3, respectively). Mollugo verticillata was the most abundant species in the MW. The prostrate growth habit of M. verticillata coupled with its relatively small plant size points to M. verticillata being not acutely problematic. In contrast, Ipomoea species are problematic in snap bean production because they vine around the crop and can interfere with harvest. Ipomoea species also are problematic in snap bean fields in the U.S. Southeast (Boyhan et al. Reference Boyhan, Harrison, Sumner, Langston, Sparks, Culpepper, Hurst, MacLean, Li, Hawkins and Fonsah2013).
Portulaca oleracea was the 10th and 14th most abundant weed in the MW and NE regions, respectively (Tables 2 and 3, respectively). Early research by Vengris and Stacewicz-Sapuncakis (Reference Vengris and Stacewicz-Sapuncakis1971) showed P. oleracea emergence within 2 wk of snap bean emergence was detrimental to crop yield. Boyhan et al. (Reference Boyhan, Harrison, Sumner, Langston, Sparks, Culpepper, Hurst, MacLean, Li, Hawkins and Fonsah2013) noted the threat of P. oleracea in snap bean could be mitigated with the use of pendimethalin, trifluralin, and S-metolachlor.
Hophornbeam copperleaf (Acalypha ostryifolia Riddell) was observed in the MW and NE regions (Tables 2 and 3, respectively). The weed was found in Illinois fields with a high density of >11.5 plants m−2 on at least one-third of fields surveyed in the state (data not shown). Illinois fields had some of the latest plantings out of all surveyed (late July), and A. ostryifolia is known to germinate quickly in warmer temperatures throughout summer (Sosnoskie et al. Reference Sosnoskie, Hayes and Steckel2020). The weed also has become widespread throughout the MW because of its innate tolerance to herbicides containing dicamba, which are commonly used in crops rotated with snap bean.
Aside from D. sanguinalis, additional annual grass species most abundant in one or more regions included annual bluegrass (Poa annua L.) and foxtail species such as giant foxtail (Setaria faberi Herrm.), green foxtail [Setaria viridis (L.) P. Beauv.], and yellow foxtail [Setaria pumila (Poir.) Roem. & Schult.] in the NE and fall panicum (Panicum dichotomiflorum Michx.) in both the MW and NE. Despite the abundance of some of these grassy species, they can be easily controlled with the use of graminicides, as there are several products registered for that purpose in snap bean (CDMS 2024). However, it is important to note that the minimum time from application to harvest (PHI) for graminicides is 15 to 21 d, which means that these herbicides can only be used until the period of snap bean flowering (Anonymous 2018, 2020, 2021). Therefore, the observed residual grassy species are most probably those that emerged after the PHI, which should be monitored in the future so that later-emerging populations are not selected for.
Common groundsel (Senecio vulgaris L.) and sharppoint fluvellin [Kickxia elatine (L.) Dumort.] were additional shared species between the NW and NE. There are no previous reports on these species in snap bean production. Senecio vulgaris is a winter annual weed found throughout the NW and is adapted to wet environments (Aldrich-Markham Reference Aldrich-Markham1994; Washington State Noxious Weed Control Board 2024). Kickxia elatine has been reported as a problematic weed in grass seed production areas of Oregon and is tolerant to many herbicides (Reference Curtis, Hinds-Cook, Hulting and Mallory-SmithCurtis et al. n.d.). Snap bean production in the NW is intensively irrigated and commonly rotated with grass seed crops (Pavlovic Reference Pavlovic2024), perhaps explaining the persistence of these weed species in snap bean.
Overall, many species that were previously reported in research and extension publications as being problematic in snap bean were also observed in the current surveys as well. Even though some of these reports are decades old, it still demonstrates that certain species or species-groups have been and continue to be problematic in snap bean. Most notable examples are C. album (Odero and Wright Reference Odero and Wright2018; Peachey Reference Peachey2019; Talbert et al. Reference Talbert, Schmidt, Burgos, Johnson and Curless1997; Van Wychen Reference Van Wychen2022; Vencill et al. Reference Vencill, Wilson and Hines1990; Wilson and Hines Reference Wilson and Hines1987) and Amaranthus spp. (Aguyoh and Masiunas Reference Aguyoh and Masiunas2003b; Lugo et al. Reference Lugo, Gonzalez and Talbert1995). Other noteworthy examples are A. artemisiifolia in the NE (Evanylo and Zehnder Reference Evanylo and Zehnder1989; Bradley et al. Reference Bradley, Bellinder and Brainard2007), P. dichotomiflorum and Setaria spp. in the NE (Teasdale and Frank 1982,Reference Teasdale and Frank1983), S. physalifolium in the NW (Peachey et al. Reference Peachey, William and Mallory-Smith2004), and D. sanguinalis in the MW (Aguyoh and Masiunas Reference Aguyoh and Masiunas2003a). Other species or species-groups seem to have been more of a problem in the past before the introduction of certain herbicide active ingredients or management practices, as they were not so abundant in the survey. These are species or species-groups such as common cocklebur (Xanthium strumarium L.) (Neary and Majek Reference Neary and Majek1990), Carolina horsenettle (Solanum carolinense L.) (Frank Reference Frank1990), Cyperus spp. (Boyhan et al. Reference Boyhan, Harrison, Sumner, Langston, Sparks, Culpepper, Hurst, MacLean, Li, Hawkins and Fonsah2013; William and Warren Reference William and Warren1975), and R. raphanistrum (Boyhan et al. Reference Boyhan, Harrison, Sumner, Langston, Sparks, Culpepper, Hurst, MacLean, Li, Hawkins and Fonsah2013). However, more importantly, the surveys also found several species or species-groups that have not been reported in snap bean previously but could be problematic in the future, such as K. elatine in the NW, A. ostryifolia in the MW, and annual bluegrass in the NE.
Relationship between Weed Density and Weed Cover
A machine learning algorithm, random forest, was used to determine the weed species or species-group that best predicted weed cover observed at crop harvest. The fitted random forest model had a pseudo-R2 value of 0.60 (±0.01) and an accuracy of 78.3%. Several species densities were strongly associated with predicting weed cover (Figure 5). Chenopodium album was the most important predictor, with 100% permutation importance. Other important predictors included D. sanguinalis (70.7%), M. verticillata (46.8%), I. hederacea (43.2%), amaranth species (35.3%), and A. ostryifolia (34.9%). Other species had a permutation importance of <22%.
In all cases, higher weed density resulted in greater predicted weed cover (Figure 6). This is consistent with density–cover (density–biomass) relationships observed by others (Hardwick and Andrews Reference Hardwick and Andrews1983; Röttgermann et al. Reference Röttgermann, Steinlein and Beyschlag2000; Weisberger et al. Reference Weisberger, Nichols and Liebman2019). Chenopodium album, D. sanguinalis, M. verticillata, I. hederacea, Amaranthus spp., and A. ostryifolia cause problems at harvest and postharvest. Above an undefined threshold, their infestation can complicate and slow snap bean harvest. Even lower infestations can result in weed organs harvested with snap bean; their removal in the processing plant can be difficult and expensive (R Pequinot, personal communication).
Figure 6. Partial dependence plots of the marginal effect of the weed species with the greatest importance in predicting mean field weed cover in the fitted random forest model.
This research documents the first quantitative report on the weed community escaping the weed management system in U.S. snap bean production. At crop harvest, weeds were present in approximately 95% of the surveyed fields. A total of 109 species or species groups were observed in 358 fields from Oregon to Delaware. In most cases, one to three species dominated the residual weed community. It was not uncommon to have >10 weeds m−2 with a weed canopy covering >5% of the field’s surface area. Some of the most abundant and problematic species or species-groups escaping control included amaranth species (such as A. hybridus, A. palmeri, A. retroflexus, and A. tuberculatus), C. album, D. sanguinalis, and I. hederacea. Acalypha ostryifolia appears to be emerging as a problematic species in the MW region. This is the first report of K. elatine routinely escaping control in snap bean, particularly in the NW. Beyond crop losses due to weed interference, the weed canopy at harvest risks contaminating snap bean products with foreign material. Our modeling suggests the residual weed canopy is dominated by C. album, D. sanguinalis, M. verticillata, I. hederacea, Amaranthus spp., and A. ostryifolia. All these species or species-groups have been identified by processors as problematic with snap bean harvest and processing.
Acknowledgments
We would like to thank students and employees of the Marty Williams Lab who assisted with data collection, including Nicholas Hausman, Christopher Landau, Jim Moody, Ana Saballos, and Yudai Takenaka. The authors also deeply appreciate the vegetable processors and farmers who participated in this research. Mention of a trademark, proprietary product, or vendor does not constitute a guarantee or warranty of the product by the U.S. Department of Agriculture and does not imply its approval to the exclusion of other products or vendors that also may be suitable.
Funding statement
This research was supported by U.S. Department of Agriculture– Agricultural Research Service Research Project No. 5012-12220-010-000D.
Competing interests
The authors declare no conflicts of interest.
Open access
