Image‐based thresholds for weeds in maize fields

• Published in: Weed research 2015-02, Vol.55 (1), p.26-33
• Main Authors: Ali, A, Streibig, J C, Christensen, S, Andreasen, C, Kudsk, Per
• Format: Article
• Language: English
• Subjects: Aquatic plants; Capsella bursa-pastoris; Chenopodium album; Cirsium arvense; Competition; Corn; crops; decision making; economic threshold; Economics; field experimentation; Field tests; Growing season; herbicide application; Herbicides; image analysis; Lamium amplexicaule; Leaves; patch spraying; pesticide application; Poa annua; Polygonum aviculare; Polygonum persicaria; regression analysis; relative weed cover; Spraying; Stellaria media; Tripleurospermum perforatum; Veronica persica; Weed control; weed management; Weeds; yield loss prediction
• ISSN: 0043-1737; 1365-3180
• DOI: 10.1111/wre.12109

Recent development of site‐specific weed management strategies suggests patch application of herbicides to avoid their excessive use in crops. The estimation of infestation of weeds and control thresholds are important components for taking spray decisions. If weed pressure is below a certain level in some parts of the field and if late germinating weeds do not affect yield, it may not be necessary the spray such places from an economic point of view. Consequently, it makes sense to develop weed control thresholds for patch spraying, based on weed cover early in the growing season. In Danish maize field experiments conducted from 2010 to 2012, we estimated competitive ability parameters and control thresholds of naturally established weed populations in the context of decision‐making for patch spraying. The most frequent weed was Chenopodium album, accompanied by Capsella bursa‐pastoris, Cirsium arvense, Lamium amplexicaule, Tripleurospermum inodorum, Poa annua, Polygonum aviculare, Polygonum persicaria, Stellaria media and Veronica persica. Relative leaf cover of weeds was estimated using an image analysis method. The relation between relative weed leaf cover and yield loss was analysed by nonlinear regression models. The competitive ability parameters and economic thresholds were estimated from the regression models. The competitive ability of weed mixtures was influenced by the increasing proportion of large size weeds in the mixtures. There was no significant effect of weeds which survived or established after the first herbicide application, indicating that early image analysis was robust for use under these conditions.