Summary
Understanding how weed communities assemble as a function of biotic and abiotic filters and transform through time has important implications for the sustainable management of agronomic systems. In a three‐year study, we evaluated weed community responses to lucerne (Medicago sativa, perennial) vs. continuous spring wheat (Triticum aestivum, annual, CSW) and weed management practices where weeds in the CSW system were managed with three contrasting approaches (herbicide, tillage or sheep grazing). Our results indicated no differences in weed diversity between the perennial and annual crops or across the different management practices in CSW. However, there were differences in weed community composition. Lucerne, with the exception of the establishing year, impeded the growth and reproduction of several annual weeds, including Amaranthus retroflexus, Thlaspi arvense, Lamium amplexicaule and Chenopodium album, but favoured perennial broad‐leaved weeds such as Taraxacum officinale and Cirsium arvense. The replacement of herbicide treatments in pre‐plant and post‐harvest in CSW with soil tillage or sheep grazing selected for different weed communities beyond the second year of establishment. The weed species driving the differences in CSW systems were Androsace occidentalis, more common in CSW managed chemically; Asperugo procumbens, more common in CSW managed with tillage; and T. officinale and Lactuca serriola, more common in CSW managed with sheep grazing. Understanding how cropping systems modify weed communities is a necessary step to shift from reactive weed control programmes to predictive management strategies.