The Prediction of Distribution of the Invasive Fallopia Taxa in Slovakia

Abstract

Invasive species are now considered the second biggest threat for biodiversity and have adverse environmental, economic and social impacts. Understanding its spatial distribution and dynamics is crucial for the development of tools for large-scale mapping, monitoring and management. The aim of this study was to predict the distribution of invasive Fallopia taxa in Slovakia and to identify the most important predictors of spreading of these species. We designed models of species distribution for invasive species of Fallopia—Fallopia japonica—Japanese knotweed, Fallopia sachalinensis—Sakhalin knotweed and their hybrid Fallopia × bohemica—Czech knotweed. We designed 12 models—generalized linear model (GLM), generalized additive model (GAM), classification and regression trees (CART), boosted regression trees (BRT), multivariate adaptive regression spline (MARS), random forests (RF), support vector machine (SVM), artificial neural networks (ANN), maximum entropy (Maxent), penalized maximum likelihood GLM (GLMNET), domain, and radial basis function network (RBF). The accuracy of the models was evaluated using occurrence data for the presence and absence of species. The final simplified logistic regression model showed the three most important prediction variables lead by distances from roads and rails, then type of soil and distances from water bodies. The probability of invasive Fallopia species occurrence was evaluated using Pearson’s chi-squared test (χ21). It significantly decreases with increasing distance from transport lines (χ21 = 118.85, p < 0.001) and depends on soil type (χ21 = 49.56, p < 0.001) and the distance from the water, where increasing the distance decrease the probability (χ21 = 8.95, p = 0.003).

Detecting and distinguishing between apicultural plants using UAV multispectral imaging

“…Some include the Random Forest (RF) ( Dmitriev et al, 2022 ; Große-Stoltenberg et al, 2016 ; Hill et al, 2017 ; Kattenborn et al, 2019 ; Sheffield et al, 2022 ; Singh & Singh, 2022 ), Classification and Regression Trees (CART) ( Fariz et al, 2022 ; Ishak et al, 2008 ; Raj & Sharma, 2022 ; Traganos et al, 2018 ), Support Vector Machine (SVM) ( Forster et al, 2017 ; Paz-Kagan et al, 2019 ; Skowronek, Asner & Feilhauer, 2017 ), Mahalanobis Minimum Distance (MMD) ( Sampedro & Mena, 2018 , Yang & Everitt, 2010 ) and Gradient Tree Boost ( Sujud et al, 2021 ). For comparative purposes, the majority of the studies have used multiple supervised classification models ( Arasumani et al, 2021 ; Gašparovičová, Ševčík & David, 2022 ; Zhu et al, 2022 ) such as Support Vector Machines (SVMs), Artificial Neural Network (ANN), Gradient Tree Boost (GTB) and Random Forest (RF) classifiers. They have been widely applied for the detection and identification of plants, in combination with the use of UAV high-resolution aerial images with ( Barrero & Perdomo, 2018 ; Bolch, Hestir & Khanna, 2021 ; Pretorius & Pretorius, 2015 ).…”