Invasive Barberry in Middlebury, VT

As temperatures rise with the increasingly harmful effects of climate change, states like Vermont become more vulnerable to the spread of invasive species. Japanese barberry (Berberis thunbergii) is one of the top invasive plant species in Vermont and is projected to have prolific population growth as the climate warms in the state. They were first brought to the US in 1875 and became a popular landscaping plant in the early 1920s as an alternative to the invasive European barberry (B. vulgaris) brought over by early European settlers. By the 1970s however, the plant was seemed as a problematic invasive species, not only displacing native species but also reducing wildlife forage, serving as a habitat for ticks and changing soil properties which may enhance the likelihood of other plant invasions.

In order to evade these harmful impacts brought on by the Japanese barberry, using cost surface analyses (CSA) through GIS is a way we can identify priority areas for removal of the plant. As a class at Middlebury College looking at the applications of GIS and natural resource management, we each came up with different criteria to analyze the least cost paths and cost surfaces. These least-cost paths represent the most likely path the Japanese barberry would spread from where the plant was first sighted in Brandon, VT and a place it was sighted in Wright Park in Middlebury, VT. These paths were derived from cost surfaces we created based on weighting factors such as frequency of flooding and elevation, where the higher the value in the cost surface, the less likely Japanese barberry would be present. 

As you can see the cost paths varied depending on the way each person weighted different criteria (methodology for my cost path is explained further down below). In thinking about how we could start tackling the issue of these invasive species, looking specifically at land that was managed by the Middlebury Area Land Trust (MALT) was a way to start to think about how existing managerial infrastructure could be enhanced to eradicate the invasive barberry plants. Looking at the least-cost paths above, all of the paths crossed through land managed by MALT or had the Trail Around Middlebury (TAM) managed by MALT run through it.

From my cost surface you can see that Chipman Hill Park is the most likely to have Japanese barberry proliferate. Seeing as many of the cost paths passed through Chipman Hill Park before reaching Wright park as well, it could be strategic to focus on doing further research in Chipman Hill Park and focus efforts on the areas where my cost path passes through through that park. While Chipman Hill Park is not owned by MALT, the TAM  that is managed by them runs through the park.

Methodology and Limitations

However, while I did use some field data to ground truth my analysis, more extensive data would need to be collected to more confidently use these cost surfaces and paths as reliable ways to target invasive Japanese barberry. 

To create my cost surface, I combined scores of 7 different categories: elevation, distance from road, presence of forest edges, frequency of floods, frost, land use and land cover (LULC) and hydrologic soil group. Separate raster layers were scored as indicated below, with research by Silander and Klepis (1999) used to determine the scoring. The lower the score, the less cost and therefore the more likely Japanese barberry would be present. 

Screen Shot 2017-04-25 at 1.50.51 PM.png

To ground truth my analysis, I teamed up with two other classmates and we collected field data from Wright Park and Chipman Hill Park. We focused on testing whether the distance form the road actually affected the presence of barberry (there was no distinction made between Japanese and European barberry). To expand the quantity of data we had, I also combined our data with the rest of the data collected by the class. I then randomly generated data points and selected points closest to them to generate the map below. 


From this, I was able to compare the percentage of data points that had barberry between distance bands to produce the graph below. While the trendline suggested a slight increase in presence of barberry further away from the road center-line, it was not a statistically significant regression as the r-squared value was 0.0145 (the closer to 1, the more statistically significant it is).


Considering that I had split the distance form the road into 2 categories, with 200 m form the road centerline being 0 and beyond that being 1, I also compared percentages in both categories. The pie chart below shows how close the percentages are, bringing the scoring of this aspect of the cost surface into question.

My investigation into the relationship between the distance from the road and the presence of barberry showed how important ground truthing is and highlights the need for organizations like MALT to use GIS as a way to target areas, but also to consistently ground truth the data so as to more effectively enact policy and programming. It would be useful to look at the methodology of the various cost surfaces that my class produced to create the cost paths in the first map in order to evaluate the accuracy of the maps and keep in mind the importance of field research despite the numerous capabilities of GIS software.

Nicole ChengComment