In this research, the deep-learning optimizers Adagrad, AdaDelta, Adaptive Moment Estimation (Adam), and Stochastic Gradient Descent (SGD) were applied to the deep convolutional neural networks AlexNet, GoogLeNet, VGGNet, and ResNet that were trained to recognize weeds among alfalfa using photographic images taken at 200×200, 400×400, 600×600, and 800×800 pixels. An increase in the image sizes reduced the classification accuracy of all neural networks. The neural networks that were trained with images of 200×200 pixels resulted in better classification accuracy than the other image sizes investigated here. The optimizers AlexNet and GoogLeNet trained with AdaDelta and SGD outperformed the Adagrad and Adam optimizers; VGGNet trained with AdaDelta outperformed Adagrad, Adam, and SGD; and ResNet trained with AdaDelta and Adagrad outperformed the Adam and SGD optimizers. When the neural networks were trained with the best-performing input image size (200×200 pixels) and the best-performing deep learning optimizer, VGGNet was the most effective neural network, with high precision and recall values (≥0.99) when validation and testing datasets were used. Alternatively, ResNet was the least effective neural network in its ability to classify images containing weeds. However, there was no difference among the different neural networks in their ability to differentiate between broadleaf and grass weeds. The neural networks discussed herein may be used for scouting weed infestations in alfalfa and further integrated into the machine vision subsystem of smart sprayers for site-specific weed control.

Field dodder is an important weed in alfalfa grown for seed, and stringent control is required to keep the alfalfa seed from becoming contaminated with dodder seed. Pendimethalin has been the primary tool used to control dodder in alfalfa seed production for more than 25 yr. Flumioxazin was recently registered in alfalfa seed and forage crops, but its activity on field dodder was unknown. Control of field dodder with flumioxazin and pendimethalin was compared in greenhouse trials in a loamy sand soil. The number of emerged dodder seedlings able to twine on a simulated host were counted weekly for a 4-wk period following herbicide application. Flumioxazin applied at the alfalfa field use rate of 0.14 kg ai ha−1 completely controlled dodder over the initial 4-wk period. Dodder either failed to emerge or emerged and died without twining on the simulated host in the 4-wk period. When dodder was reseeded after 4 wk in flumioxazin-treated soil, the number of twined dodder seedlings was reduced by 56% in one trial but was unaffected in a second trial compared with nontreated checks. In the initial 4-wk period following herbicide application, flumioxazin controlled field dodder similar to pendimethalin applied at 2.2 and 4.4 kg ai ha−1 in both trials. However, after additional dodder seed was planted at 4 wk after treatment, dodder was suppressed more by pendimethalin than flumioxazin in the additional 4-wk period. Flumioxazin offers alfalfa seed and forage producers a new mode of action to manage early-emerging field dodder.

Velvetleaf is difficult to control in corn and soybean and the seed can persist in soil for many years. Seven cultural and tillage practices were established in 1974 on a site heavily infested with velvetleaf to determine the time required to eradicate velvetleaf seed from the soil. A rapid decline in velvetleaf seed population in the top 23 cm of soil occurred during the first 5 yr of this study. In the fifth year, the chemical fallow and continuous alfalfa treatments had 37 and 56% of the original velvetleaf seed population remaining, respectively. In the 17th year, soils in these treatments that had received no tillage since study initiation still contained 15 and 25% of the original velvetleaf seed population, respectively. Systems involving moldboard plowing with continuous-tillage fallow, continuous cropping of corn or oat, or an annual corn and soybean rotation had a more rapid decline in the velvetleaf seed population in soil compared to the chemical fallow and continuous alfalfa treatment. After 17 yr, soil in any system that had received at least one moldboard plowing per year still contained 1 to 3 million velvetleaf seed ha−1, which is only 0.8 to 2.5% of the initial viable seed population. Nearly 100% of the seed remaining in the soil in the 17th year for all treatments was still viable.

Field research from 1986 to 1988 evaluated control of alfalfa, dandelion, orchardgrass, and annual weeds in corn with sequential herbicide treatments. Glyphosate applied the previous fall controlled 95% or more of the dandelion and orchardgrass before postemergence treatments. Atrazine applied early preplant or with paraquat the day before planting only partially controlled the perennial species. Atrazine applied postemergence usually controlled vegetation escaping initial treatment except orchardgrass following the atrazine treatments. Adding tridiphane increased weed control with atrazine or cyanazine applied postemergence only when fall panicum was present. Herbicides applied postemergence increased corn yield compared to the initial treatments alone, and few differences among postemergence treatments were observed. Sequential treatments which included glyphosate applied in the fall controlled 82% or more of each perennial species before the next growing season. Orchardgrass control was 81% or less the following spring when the initial treatment was atrazine or atrazine plus paraquat.

Chlorsulfuron was applied to wheat at rates of 0, 10, 20, and 40 g ai ha-1 in three locations at Lethbridge and Brooks between 1980 and 1984. The soil pH and organic matter contents at Lethbridge and Brooks were 8.0 and 2.1% and 7.4 and 3.0%, respectively. Rotational crops were seeded for up to 7 yr after chlorsulfuron application. The criteria used to indicate when susceptible crops will grow safely were no significant yield loss at the 40 g ha-1 rate and the magnitude of the upper confidence limit for this difference. The following times after application before crop yields were not affected were required at Lethbridge: barley, 2 yr; canola, 3 yr; peas, beans, 4 yr; flax and potatoes, 5 yr; alfalfa and sugarbeets, 6 yr; and lentils at least 7 yr. The effect of chlorsulfuron residues in the soil on rotational crops appeared to be similar at Brooks and at Lethbridge. Estimated chlorsulfuron concentrations were related to the tolerance of rotational crops.

Quinclorac was registered for weed control in wheat (Triticum spp.) for western Canada in 1997. Residues from quinclorac may persist in the soil and may damage following crops; therefore, field and growth chamber experiments were conducted to determine the tolerance of several following rotational crops. Cereals and bromegrass (Bromus biebersteinii) were sufficiently tolerant that they could be seeded within 16 d of quinclorac application without risk of injury. At the other extreme, marketable and total potato (Solanum tuberosum) yields were reduced by quinclorac on irrigated land 1 yr after application. Growth chamber experiments were used to rank crops in order of their tolerance of quinclorac residues and to compare the tolerance of crops that were grown in the field with additional crops. Quinclorac injured several legume and oilseed crops when the crops were seeded immediately after application, but quinclorac did not reduce the dry matter yield of two of the most sensitive legumes, faba bean (Vicia faba) and alfalfa (Medicago sativa), 1 yr after application on irrigated land. However, based on a previous study, one can conclude that injury to these crops may occur in the field under drought conditions in rain-fed agricultural systems.

Two experiments were conducted near Scottsbluff, NE, to evaluate the efficacy of fall-and spring-applied herbicides for downy brome control in established alfalfa. Downy brome was effectively removed from established alfalfa, and alfalfa yield increased with fall applications of hexazinone, metribuzin, pronamide, and terbacil. Regression analysis indicated a linear relationship between alfalfa yield and downy brome biomass. Glyphosate or paraquat suppressed downy brome when applied to dormant alfalfa in the spring. If glyphosate or paraquat application was delayed until after alfalfa had resumed spring growth, injury was observed. Alfalfa yield did not increase following spring applications of glyphosate or paraquat.

R230, an isolate of Rhizoctonia solani anastomosis group 5 from leafy spurge, caused root and crown rots on leafy spurge when inoculated via soil in greenhouse experiments. The fungus grew through soil over a wide range of matric potentials (–0.01 to −1 MPa). When applied to leafy spurge foliage, the fungus blighted young adventitious shoots, but not mature stems. Foliar infection required high relative humidity. Mycelial growth on plant surfaces stopped at below 92% relative humidity. Pathogenicity of R230 was not limited to leafy spurge. The fungus caused stem and foliar lesions on all crops tested (alfalfa, smooth bromegrass, corn, Kentucky bluegrass, soybean, tall fescue, and wheat) when plants were kept under high humidity. It reduced seedling emergence and survival in soybean, Kentucky bluegrass, and tall fescue. R230 also caused a foliar blight on spotted spurge. The potential for R230 to damage crops will limit application of the fungus to noncultivated areas and to sites where it is endemic.

‘Vernal’ alfalfa response was evaluated when chlorsulfuron, thifensulfuron, 2,4-D, glyphosate, bromoxynil, and selected combinations of those herbicides were applied at rates simulating spray drift during the fourth trifoliolate leaf stage following the first cutting in 1990 and 1991. The order of phytotoxicity was 2,4-D > chlorsulfuron > thifensulfuron > glyphosate > bromoxynil. By the end of each growing season, alfalfa had recovered from injury caused by all herbicides except the highest rates of 2,4-D and 2,4-D plus glyphosate. The alfalfa stand was reduced only by 2,4-D and 2,4-D plus glyphosate. All herbicides caused characteristic symptoms, but some specific symptoms were similar among different herbicides or resembled symptoms caused by disease, mineral imbalance, and adverse weather conditions.

Field experiments were established to evaluate the effect of herbicide treatments on control of tall fescue before spring or fall no-till alfalfa planting and to determine the effect of fescue control on alfalfa seedling establishment, forage yield, and forage quality. No treatment provided significantly greater tall fescue control than the standard two applications of paraquat at 13- to 21-day intervals. Fescue control values with this treatment ranged from 86 to 99% over experiments and were associated with maximum values for alfalfa ground-cover and alfalfa yield percentage parameters, 52 to 83% and 64 to 96%, respectively. Glyphosate and SC-0224 at 2.2 kg/ha and greater generally gave tall fescue control equivalent to the paraquat treatment in both spring and fall seedings as did sethoxydim at 0.6 kg/ha in spring seedings. Applications of fluazifop or HOE-0661 did not result in satisfactory tall fescue control or alfalfa establishment. Adding nonionic surfactant to glyphosate or SC-0224 did not increase tall fescue control. There was a direct linear relationship between tall fescue control and both alfalfa seedling establishment and yield percentage.

Trifluralin selectively controls dodder in alfalfa. However, its soil activity dissipates allowing some late season dodder germination. Thiazopyr applied PRE was also found to control dodder selectively in alfalfa. Thiazopyr was more effective when applied in February than in November. Thiazopyr did not control dodder longer into the season than trifluralin in field tests, however it was more effective in greenhouse experiments. The effect of thiazopyr on dodder shoot tips was similar to that of trifluralin, in that both herbicides produced bulbous cells with thickened, globular nuclei, typical of cells treated with herbicides affecting microtubules.

Glyphosate and SC-0224 at 75 or 150 g ae/ha, applied to alfalfa foliage after field dodder and largeseed dodder were parasitically well established, controlled dodder selectively. Glyphosate applied at any time over a range of developmental stages stopped the growth of dodder. Control was best when the herbicides were applied when alfalfa was 20 to 30 cm tall, when maximum length of dodder shoots was 30 to 60 cm, and while host and parasite were both vegetative and growing vigorously. Later application allowed dodder to injure alfalfa before being controlled, and earlier application sometimes allowed dodder to recover and suppress alfalfa before the hay was cut.

This study was conducted to compare foliar absorption and translocation of dicamba applied to plants as an aqueous solution or as dicamba-treated soil deposits. When applied as an aqueous solution, 65 to 95% of 14C-dicamba was absorbed by pea, alfalfa, and grape; whereas, limited absorption (0.4 to 4.7%) occurred from dicamba-treated soil. Based on dose-response evaluation, the low levels absorbed from treated soil are unlikely to cause crop damage. 14C-dicamba translocation was more than 80% of absorbed 14C in pea, alfalfa, and grape. Acropetal translocation exceeded basipetal translocation.

A computer model which selects least cost herbicide programs given a minimum desired level of weed control could provide growers with economical weed management options. Using an integer programming approach, a herbicide selection model was developed for corn production under Pennsylvania conditions. Models for three rotations (corn-soybean, corn-corn, and corn-alfalfa) under three tillage systems (conventional tillage, reduced tillage, and no-till) that evaluated 21 soil-applied and 13 postemergence herbicide options for 24 weeds were developed. Each model minimizes the cost of a herbicide program subject to a desired level of weed control. By selecting the weed species to be controlled and the level of control desired, customized herbicide programs can be generated. The models can also be used to evaluate the cost of changing the level of control desired for an individual weed species or set of weeds.

Field experiments were conducted during 1981 to 1984 to evaluate paraquat for weed control in alfalfa established for less than one year. Treatments were applied in March to alfalfa up to 4 cm tall and/or after cuttings to alfalfa up to 10 cm tall. Weeds were up to 20 cm tall. Paraquat applied in March controlled common chickweed and shepherd's-purse in some experiments. It did not control birdsrape mustard when applied after cuttings. Paraquat suppressed growth of perennial broadleaf weeds and grasses and reduced pressure from annual grasses in some experiments. March treatments generally did not increase, but sometimes decreased alfalfa yields. Paraquat applied after cuttings did not affect alfalfa yields in the majority of the experiments.

This review considers the development of herbicide-resistant weed biotypes in Australia. Biotypes of the important annual weed species, capeweed, wall barley, and hare barley are resistant to the bipyridylium herbicides paraquat and diquat. These resistant biotypes developed on a small number of alfalfa fields that have a long history of paraquat and diquat use within a distinct geographical area in central western Victoria. The resistant biotypes are controlled by alternative herbicides and pose little practical concern. Some populations of wild oat are resistant to the methyl ester of diclofop. Of greatest concern is the development of cross resistance in biotypes of rigid ryegrass to aryloxyphenoxypropionate, cyclohexanedione, sulfonylurea, and dinitroaniline herbicides. The cross-resistant rigid ryegrass infests crops and pastures at widely divergent locales throughout the cropping zones of southern Australia. The options for control of cross-resistant rigid ryegrass by herbicides are limited. A biotype of rigid ryegrass on railway tracks treated for 10 yr with amitrole plus atrazine has resistance to amitrole and atrazine and other triazine, triazinone, and phenylurea herbicides. Management tactics for cross resistance are discussed.

Newly seeded alfalfa tolerated glyphosate applied from the 8-trifoliolate leaf stage (17 cm tall) to the pre-bud stage (34 cm tall) at 75 or 150 g ae ha-1. Injury symptoms were moderate and temporary. Glyphosate at 300 g ha-1 suppressed alfalfa severely and killed 30% of the plants when applied at the smallest (3-trifoliolate) stage of alfalfa growth. Alfalfa responded similarly to glyphosate and SC-0224.

Ten herbicides were evaluated for poison-hemlock control in established alfalfa in Utah. Hexazinone, metribuzin, and terbacil controlled poison-hemlock 90% or more but diuron and pronamide did not. Either glyphosate or 2,4-D applied in May at 1.1 kg ae ha-1 controlled poison-hemlock 97 to 98% compared to 60 to 75% with 2,4-DB at that rate. Dicamba and bromoxynil did not control poison-hemlock.

An accurate method is described for evaluating the relative effectiveness of isolates of Fusarium avenaceum as potential biological control agents for weeds such as spotted knapweed and wild oat. The method uses seeds that have been disinfected, placed in tubes containing water agar, and challenged with various isolates of F. avenaceum. Evaluatory data on the pathogenicity of the fungi are obtained after 14 d, when plants and fungi both are grown on water agar in the lab. Commonly, potential biocontrol agents are tested on seeds placed in the soil, but up to 60 d may be required for results. However, comparative studies using the soil test vs. the water agar test indicated that the water agar tests are giving comparable results to those from greenhouse tests but in a shorter time (14 vs. 60 d). Also, both water agar and soil tests were done with a number of crop plants including wheat, barley, alfalfa, and wild oat with comparable results. Thus, this method has the potential to screen many pathogens that may have potential biocontrol activities in a short time.

EPTC applied at the time of ‘Saranac AR’ alfalfa establishment resulted in higher levels of crop injury from the octanoic acid ester of bromoxynil applied postemergence compared to alfalfa in areas not treated with EPTC. EPTC applied at 1.7 or 3.4 kg ai ha-1 did not injure alfalfa. At Rock Springs, PA, 0.54 kg ai ha-1 bromoxynil caused 28, 43, and 55% foliar injury in areas treated with 0, 1.7, and 3.4 kg ai ha-1 EPTC, respectively. A similar interaction was observed at Landisville, PA, although the level of injury due to bromoxynil was less severe. EPTC also predisposed ‘WL 316’ alfalfa to bromoxynil injury in greenhouse experiments.