Determining representative pseudo-absences for invasive plant distribution modeling based on geographic similarity

Abstract

Introduction: The use of pseudo-absence data constrained by environmental conditions can facilitate potential distribution predictions of invasive species. However, pseudo-absence data generated by existing methods are usually not representative because the relationship between the presence and pseudo-absence points is either simplistic or neglected. This could under or overestimate the potential distribution of invasive species. Methods: To address this deficiency, this study proposes a new method for obtaining pseudo-absence data based on geographic similarities. First, the reliability of pseudo-absences was quantified based on the geographic similarity to the occurrence of species. Subsequently, a representative pseudo-absence reliability threshold interval was determined. Finally, different pseudo-absence acquisition methods were assessed by combining virtual species with a real invasive species. Results: The analysis demonstrated that the geographic similarity method can improve model accuracy and achieve a more realistic distribution compared with the traditional method of sampling for pseudo-absence data. Discussion: This result indicates that the pseudo-absence data obtained using the geographic similarity approach were more representative. Our study provides valuable insights into improving invasive plant distribution predictions by considering the geographical relationships between species occurrences and the surrounding environments.