Detecting invasive species using remote sensing techniques is becoming increasingly popular. However, there are only a few species that have a distinct spectral signature and unique phenology which allows for better accuracy for detection and monitoring.  Acacia dealbata Link (silver wattle) is an invasive species in south-central Chile native to Australia, its intense yellow flowers in the middle of winter, provide a distinct pattern in color aerial photography.  Acacia dealbata is associated to roads and other human disturbances and riparian habitats, and invades most areas in the coastal range of south central Chile. We are studying the species across landscape and regional scales to detect invasion processes. We want to 1) assess current patterns of invasion, 2) detect changes across time and 3) improve methods for detection of other similar invasive species. We have used aerial photos (1:20,000) from 2003 and 2007. We have randomly selected 9 by 9 km quadrats to detect spatial patterns. We have built a vector grid consisting of 30*30 m, 90*90 m or 150*150m.  In each pixel, using visual photointerpretation, presence of A. dealbata was recorded when more than 5% of the pixel was occupied by the species. To increase our understanding of the invasion of A. dealbata, we analyzed the relationship between the mapped infestations with a number of data layers (e.g., roads, rivers, land use). 

Results/Conclusions

We have found that the species is significantly associated with rivers but less so with roads (logistic regression, p<0.01).  There is significant aggregation of the species from the stand to the landscape scale (Modified Ripley’s K, L-function). This may indicate that the species has a continuous distribution over the landscape once it establishes itself in a new location and that the species is still spreading over the landscape and that limitations for establishment may be more related to propagule pressure than to environmental conditions. By contrasting 2003 and 2007 photos we have detected temporal changes at the smaller scales. We are now focusing in modeling the major drivers for invasion expansion across scales to be able to predict future scenarios. The proposed method is simple and can be used where other more expensive remote sensing techniques are not available. Funding provided by grants Fondecyt 1070488 and ICM P05-002.