Evaluating habitat suitability and spectral heterogeneity models to predict weed species presence

Sammanfattning: The aim of this study is to analyze whether the presence of weed species resistant to herbicides is related to landscape heterogeneity and environmental variables. C3 and C4 weed species, such as Chenopodium album, Conyza canadensis, Amaranthus retroflexus, Papaver rhoeas, Sorghum halepense, Echinochloa crus-galli and Lolium rigidum, are studied in two differing bioclimatic countries, Greece and Germany. Given the difficulties of field-based data collection, the use of remote sensing for estimating spatial patterns is as a powerful tool. To compare species presence with spatial heterogeneity, the Normal Difference Vegetation Index (NDVI) was used. A number of tools were applied such as regression models, Ordinary Least Square (OLS) and Logistic regression, Rank - abundance diagrams and Variograms of NDVI values. Moreover, to relate species distribution with environmental and bioclimatic variables Habitat Suitability (HS) models were produced. To make predictions of species future distribution, based on bioclimatic data, GARP models were used. Landscape area around species is highly heterogeneous. The species studied are related with low vegetated areas and are neither generalists nor specialists in their environment. The core factors affecting species distribution are mean temperature, and secondarily annual precipitation, in Greece, whereas altitude and bioclimatic conditions, in Germany. Weeds tend to move to higher altitudes, minimizing their population size, in future climatic conditions. However, C3 summer annual weeds, in Germany, and C3 winter annual weeds, in Greece, seem to be tolerant to climatic changes, and expand to new territories. The response of species range due to climatic changes, seems to be species-specific. Further study is needed to come up with prediction models of weed species future distribution