Abstract
In recent years, an increasing number of distribution maps of invasive alien plant species (IAPS) have been published using different machine learning algorithms (MLAs). However, for designing spatially explicit management strategies, distribution maps should include information on the local cover/abundance of the IAPS. This study compares the performances of five MLAs: gradient boosting machine in two different implementations, random forest, support vector machine and deep learning neural network, one ensemble model and a generalized linear model; thereby identifying the best‐performing ones in mapping the fractional cover/abundance and distribution of IPAS, in this case called Prosopis juliflora (SW. DC.). Field level Prosopis cover and spatial datasets of seventeen biophysical and anthropogenic variables were collected, processed, and used to train and validate the algorithms so as to generate fractional cover maps of Prosopis in the dryland ecosystem of the Afar Region, Ethiopia. Out of the seven tested algorithms, random forest performed the best with an accuracy of 92% and sensitivity and specificity >0.89. The next best‐performing algorithms were the ensemble model and gradient boosting machine with an accuracy of 89% and 88%, respectively. The other tested algorithms achieved comparably low performances. The strong explanatory variables for Prosopis distributions in all models were NDVI, elevation, distance to villages and distance to rivers; rainfall, temperature, near‐infrared and red reflectance, whereas topographic variables, except for elevation, did not contribute much to the current distribution of Prosopis. According to the random forest model, a total of 1.173 million ha (12.33% of the study region) was found to be invaded by Prosopis to varying degrees of cover. Our findings demonstrate that MLAs can be successfully used to develop fractional cover maps of plant species, particularly IAPS so as to design targeted and spatially explicit management strategies.
Highlights
In the last 20 years, many studies have attempted to accurately detect the spatial extent of invasive alien plant species (IAPS) to map their spread over time or model their potential invasion area
We set out to compare the performances of five machine learning algorithms (MLAs), an ensemble model and a generalized linear model
We chose five MLAs: two different implementations of gradient boosting machine (GBM and GBM‐BRT), random forest (RF), support vector machine (SVM), and deep neural network (DNN), an ensemble model composed of the four best‐performing tested algorithms, and a generalized linear model (GLM) for comparison reasons
Summary
In the last 20 years, many studies have attempted to accurately detect the spatial extent of invasive alien plant species (IAPS) to map their spread over time or model their potential invasion area. More promising mapping of IAPS at finer fractions of cover was obtained using a combination of medium or high‐resolution satellite data and powerful machine learning classification algorithms (Ng et al, 2016; Rembold, Leonardi, Ng, Gadain, & Meroni, 2015) Such fine‐scaled and accurate quantification of the local fractional cover of IAPS allows understanding their impacts through cover‐impact curve analysis. We set out to compare the performances of five MLAs (gradient boosting machine implemented in two different ways, random forest, support vector machine, and deep learning neural network), an ensemble model and a generalized linear model This analysis helps identifying the best‐performing algorithm in mapping detailed fractional cover of Prosopis in the dryland ecosystem of the Afar Region, Ethiopia. The best‐performing model was used to create a Prosopis distribution and fractional cover map
Sign-in/Register to view highlights & summary Sign-in/Register to access full text options 
Free) 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call
