Purple loosestrife (Lythrum salicaria L.) is an invasive wetland plant species of Eurasian origin. Loosestrife has spread widely through much of the USA, and is widespread in the state of Minnesota. Minnesota’s lake saturated landscape makes it both extremely vulnerable to loosestrife invasion, and also an interesting system to study the ecological processes involved in the landscape level distribution of Loosestrife. The overarching aim of our study is to explore and explain the observed distribution pattern of loosestrife for the entire state of MN using available survey data. We used GIS derived land cover data and overlaid it with the observed distribution of loosestrife provided by the Minnesota DNR. Focussing on three counties with comparable loosestrife presence records, but with widely differing environmental, latitudinal, and anthropogenic effects we explored the sensitivity of different land cover types to loosestrife invasion.

Results/Conclusions

Using logistic regression, we found that the ‘open water’ land cover type had the highest probabilities (0.083 to 0.091) for being invaded by loosestrife.  The next best explanatory variable was under the broad category of ‘developed’ land cover types with a nearly consistent probability coefficient ranging between 0.041 and 0.047. However, the ‘developed’ land cover type was statistically significant for only one county. From this preliminary analysis, it appears that most of the observed distribution can be explained by presence of open water habitats like lakes. The role of anthropogenic disturbance is also apparent, as suggested by the relatively high probability coefficients found for ‘developed’ land cover types. In the next step of our analysis, we assess how well the derived probabilities of open water land cover estimate loosestrife presence for the remaining counties of Minnesota. Ultimately, we shall use the tools of network analysis to compare random networks of lake basin and open water habitats occupied by loosestrife with more ecologically realistic networks based on loosestrife dispersal and sensitivity to the spatial arrangement of land cover types.