No abstract

Weed seedling emergence is partially dependent on biotic and abiotic conditions directly surrounding the seed. When environmental conditions are appropriate, seed germination and emergence occurs. We studied the impact of seeding depth (surface, 1 to 2, 3 to 4, and 6 to 7 cm) and fluctuating soil moisture regimes (field capacity [FC]–1/3FC–FC; FC–1/6FC–FC) on percent weed emergence in a greenhouse. At FC, wild mustard and field pennycress had the greatest percent emergence when seeds were placed on or near the soil surface, whereas percent emergence of barnyardgrass and round-leaved mallow was unaffected by seeding depth. All the perennials tested had the greatest percent emergence at FC when seeds were placed near or on the soil surface, except for common milkweed which only emerged below the soil surface. When soil moisture levels fluctuated, surface seeds of barnyardgrass, catchweed bedstraw, green foxtail, wheat, and wild oat had less emergence than seeds below the soil surface; field pennycress had increased emergence when the seeds were placed on the surface; and round-leaved mallow and wild mustard emergence was unaffected by seeding depth. The emergence of curly dock, dandelion, and perennial sowthistle was unaffected by seeding depth, whereas foxtail barley and quackgrass emergence was reduced when seeds were placed on the surface and soil moisture fluctuated.

Long-term research on cover crops (CC) is needed to design optimal rotations. Winter CC shoot dry matter (DM) of rye (Secale cereale L.), legume-rye, and mustard was determined in December to February or March during the fi rst 8 yr of the Salinas Organic Cropping Systems trial focused on high-value crops in Salinas, CA. By seed weight, legume-rye included 10% rye, 35% faba (Vicia faba L.), 25% pea (Pisum sativum L.), and 15% each of common vetch (V. sativa L.) and purple vetch (V. benghalensis L.); mustard included 61% Sinapis alba L. and 39% Brassica juncea Czern. Cover crops were fall-planted at 1x and 3x seeding rates (SR); 1x SR were 90 (rye), 11 (mustard), and 140 (legume-rye) kg ha -1 . Vegetables followed CC annually. Cover crop densities ranged from 131 to 854 plants m -2 and varied by CC, SR, and year. Year, CC, and SR aff ected DM production, however, the eff ects varied across the season and interactions occurred. Averaged across years, fi nal DM was greater in rye and legume-rye (7 Mg ha -1 ) than mustard (5.6 Mg ha -1 ), and increased with SR through January. Dry matter production through the season was correlated signifi cantly with growing degree days (GDD). Legumes contributed 27% of fi nal legume-rye DM. Season-end legume DM was negatively correlated with GDD at 30 d, and legume DM in the 3x SR increased during years with frequent late-season rainfall. Seed costs per Mg of fi nal CC DM at 1x SR were approximately three times higher for legume-rye than rye and mustard.

Precision herbicide application can substantially reduce herbicide input and weed control cost in turfgrass management systems. Intelligent spot-spraying system predominantly relies on machine vision-based detectors for autonomous weed control. In this work, several deep convolutional neural networks (DCNN) were constructed for detection of dandelion (Taraxacum officinale Web.), ground ivy (Glechoma hederacea L.), and spotted spurge (Euphorbia maculata L.) growing in perennial ryegrass. When the networks were trained using a dataset containing a total of 15,486 negative (images contained perennial ryegrass with no target weeds) and 17,600 positive images (images contained target weeds), VGGNet achieved high F1 scores (≥0.9278), with high recall values (≥0.9952) for detection of E. maculata, G. hederacea, and T. officinale growing in perennial ryegrass. The F1 scores of AlexNet ranged from 0.8437 to 0.9418 and were generally lower than VGGNet at detecting E. maculata, G. hederacea, and T. officinale. GoogleNet is not an effective DCNN at detecting these weed species mainly due to the low precision values. DetectNet is an effective DCNN and achieved high F1 scores (≥0.9843) in the testing datasets for detection of T. officinale growing in perennial ryegrass. Moreover, VGGNet had the highest Matthews correlation coefficient (MCC) values, while GoogleNet had the lowest MCC values. Overall, the approach of training DCNN, particularly VGGNet and DetectNet, presents a clear path toward developing a machine vision-based decision system in smart sprayers for precision weed control in perennial ryegrass.