ABSTRACT

Long-term field experiments were conducted from 1985 through 1992 at Stoneville, MS on a Dundee sandy loam (Aerie ochraqualf) soil to evaluate the effects of four cotton (Gossypium hirsutum L.) tillage production systems on the population dynamics and seedbanks of weeds and subsequent effects on cotton growth and yield. Cotton tillage production systems included: (a) no-tillage (NT); (b) conventional tillage (CT) consisting of fall subsoiling, spring chisel plowing, double-disk incorporation of preplant herbicide, bedding, knocking down beds, and three cultivations; (c) reduced-tillage high level (RTH) which lacked double disking only; and (d) reduced-tillage low level (RTL) which lacked double disking and chisel plowing. Two herbicide levels were evaluated for each of the tillage systems. Thirty nine of the 82 weed species found in the experiment during the 8-yr period were recorded from the NT treatments only. In addition to the most weed diversity, the greatest populations per species were found in the NT treatments regardless of the herbicide input level. Overall control of broadleaf, grass, and sedge weed species was higher in both herbicide input levels for CT, RTH, and RTL when compared to the NT treatments in all 8 years. Herbicide inputs alone were ineffective in controlling all weed species in all treatments regardless of tillage production system; however, individual species weed populations were from 1.2- to 120-fold greater in the NT treatments than in the CT, RTH, and RTL production systems regardless of the herbicide input level. Pigweeds (Amaranthus sp.), prickly sida (Sida spinosa L.), purple and yellow nutsedges (Cyperus rotundus L. and C. esculentus L.), and spurges (Euphorbia sp.) populations increased the greatest over years when tillage levels decreased.

Arrowleaf sida (Sida rhombifolia L.) populations increased only in the NT treatments. Three weedy species, a small sand parsley [Ammoselinum butleri (S. Wats.) Coult. & Rose], a hispid pepperweed (Lepidium mistrinum Small), and hardgrass [Sclerochloa dura (L.) Beauv.], were recorded from NT treatments only and each of these species were new to the Mississippi flora. Soil samples were taken in 1990 and 1991 to determine weed seedbanks in each production system. Seedbanks of monocot (grass and sedge) and dicot (broadleaf) weed species were 2- to 3-fold greater in the NT treatments than in the CT, RTH, and RTL treatments regardless of the herbicide input level. Seedbanks of common purslane (Portulaca oleracea L.), pigweeds (Amaranthus sp.), and spurges (Euphorbia sp.) and several winter weed annual species, corn speedwell (Veronica arvensis L.), henbit (Lamium amplexicaule L.), shepard's purse [Capsella bursa-pastoris (L.) Medicus], and Virginia pepperweed (Lepidium virginianum L.), increased the greatest in NT treatments when compared to the CT, RTH, and RTL production systems. Seed cotton yields were not significantly different for CT, RTH, and RTL treatments at both herbicide input levels in 7 of 8 years (except 1985); however, yields were reduced in NT treatments regardless of the herbicide input level in 7 of 8 years (except 1991). Based on these data, the RTH and RTL systems for cotton production required fewer inputs without appreciable losses in cotton yields and weed control when compared to CT but in the NT production system, cotton yields where reduced and weed populations are greater and more diverse despite equivalent herbicide input levels.