98th ESA Annual Meeting (August 4 -- 9, 2013)

COS 61-9 - Global modeling of invasive Parthenium hysterophorus reveals new areas of potential invasion on multiple continents

Wednesday, August 7, 2013: 10:50 AM
L100I, Minneapolis Convention Center
Kumar P. Mainali, Section of Integrative Biology, University of Texas at Austin, Austin, TX, Abhishek Nakarmi, Environmental Science Institute, The University of Texas at Austin, Kunjithapatham Dhileepan, Invasive Plant Science, Biosecurity Queensland, Department of Employment, Economic Development & Innovation, Ecosciences Precinct, Brisbane, Qld 4001, Australia, Dan Warren, Integrative Biology, University of Texas, Austin, Austin, TX, Andrew McConnachie, PPRI Weeds Lab, Private Bag X6006, Hilton, South Africa, Lorraine Strathie, ARC-Plant Protection Research Institute, Private Bag X6006, Hilton, South Africa, Matthew P. Moskwik, Department of Integrative Biology, University of Texas, Austin, TX, Gul Hassan, Department of Weed Science, NWFP Agricultural University, Peshawar, Pakistan, Iqbal Zuberi, To be added later, Bangladesh, Debendra Karki, Central Department of Botany, Tribhuvan University, Kathmandu, Nepal, Bharat B. Shrestha, Central Department of Botan, Tribhuvan University, Kathmandu and Camille Parmesan, Marine Institute, Plymouth University, Plymouth, United Kingdom
Background/Question/Methods

Parthenium weed (Parthenium hysterophorus L., Asteraceae), a native of tropical America, is a weed of global significance.  It was first identified outside its native range in Australia in 1955 and then India in 1956. Since the 1950s, the plant has spread to many tropical and sub-tropical areas (three dozen countries) of the world.  In the invaded ranges, parthenium has substantial negative impact on crop, rangeland productivity, native community biodiversity and the health of humans and animals. Given the magnitude of invasion and intensity of impact of the weed, we attempted to determine whether there are additional global areas where it could invade and whether it’s reached equilibrium within already invaded ranges. We created 30,000 models using four species distribution methods (Random Forest, Boosted Regression Trees, Generalized Linear Model and Generalized Additive Model), 26 environmental predictors (altitude, tree canopy cover, soil moisture, proximity to road, human population density as a proxy of domestic waste, and 19 WorldClim variables), and several novel methods of selecting occurrence and background points.

Results/Conclusions

Our results indicate that the plant is still spreading and has likely not reached equilibrium in the non-native ranges as evidenced by the following: (1) many regions in East Asia, Africa and Australia where the plant has not yet invaded harbor highly suitable habitat for the weed, (2) models with background points drawn from the world were substantially worse than models with background points drawn from native ranges (a difference in area under the ROC curve [AUC] score of 10.4%). Additionally, we show that roads do not provide better habitats for the weed. Therefore, the common observation of the plant rapidly spreading along roads in Asia and Africa is reduced to conclude that roads only assist in propagule dispersal. With AUC scores of 0.85–0.90, the models yield a consensus map of habitat suitability for the plant which can be useful in making conservation plans in Asia, Africa, Australia and Europe.