Aspectos espaciales y temporales de la comunidad de malas hierbas en cultivos estivales de regadío en el centro de España

Abstract

The concept of Site-Specific Weed Management (SSWM) (using herbicides only in places where weed density is above a certain threshold) may be used in order to reduce the huge amount of herbicides used in the traditional management. For that purpose, it is required to know the places where weeds are more abundant and to determine the spatial distribution pattern of these weeds, since SSWM is justified only when weeds exhibit aggregated distribution. However, there are some aspects of SSWM with possibility of improvement. Weed detection is one of the challenges SSWM has to face and the employed scale (sampling resolution, treatment resolution, etc.) is a key factor that should be kept in mind. The main goals of this thesis are: 1) to identify the spatial structure of weed communities in corn crops in the centre of Spain at different scales; 2) to determine the possible herbicide savings when using SSWM systems based on prescription maps simulated with different types of resolutions and detection thresholds; and 3) to determine the spatio-temporal evolution of Sorghum halepense, an important perennial weed in corn and poplar short-coppices in central Spain. The results have shown that the major weed species in corn crops have an aggregated spatial pattern at different scales, from cm to plot scale. However no conclusion has been reached regarding to association/dissociation patterns between species, regardless of scale study. A relationship between plant density and soil compaction from machinery was found, in a way that, depending the species, larger or smaller density was found in the compacted area comparing to un-compacted one. With regard to herbicide savings based on prescription maps, we have concluded that the use of SSWM derives on important saving when species are aggregated and we use relatively high detection thresholds (which conform to reality of actual sensors). On the other hand, in more evenly distributed species, herbicide savings are scarce and maps errors (infestation areas not treated) increase. Hence, SSWM is not recommended in this case. Regarding to spatio-temporal evolution of S. halepense patches in corn crops, geometric models have been developed to predict the areas of expansion of this species in space and time. These models would be useful in Decision Support System for SSWM. Specifically, two geometric models, triangular and rectangular, were used each showing weaknesses and strengths. The triangular model could predict properly the expansion pattern of already in-fields patches, but not accurate in predicting the emergence of new foci. Instead, the rectangular model properly simulated the position of new foci, but maximizing the infested area of S. halepense. Therefore, the use of rectangular model assumes a more conservative strategy, with less herbicide saving but also minor errors regarding untreated infested areas. In poplar short-coppices we observed that the spread of this species is limited, especially when chemical treatments are employed. However competition with the crop could be high, especially during the first growing season. Consequently, an appropriate management of S. halepense during this initial year is highly recommended.