Modeling the Spatiotemporal Habitat Suitability Distributions of Cultural Keystone

Bidens macroptera symbolizes the change of a season, marking the transition from the rainy season to autumn, heralding the new year for Ethiopians. Despite a general understanding of its geographic regions, significant gaps remain in identifying the habitat distribution and key predictor variables of Bidens macroptera through species distribution modeling (SDM) in the context of climate change. We developed an ensemble species distribution model using 2 statistical and 3 machine learning algorithms. We collected 119 presence and pseudoabsence points to train and validate bioclimatic variables through 5-fold cross-validation with the "SDM" package in R software. We calculated the Variance Inflation Factor (VIF) to assess multicollinearity among environmental variables. Projections were made for medium (SSP 2-4.5) and extreme (SSP 5-8.5) greenhouse gas emissions for the periods of the 2050s and 2070s. The performance of the models was evaluated by Area Under the Curve (AUC) and True Skill Statistic (TSS). A weighted average threshold value at the sum of sensitivity and specificity of the TSS was used to classify habitat suitability using ArcGIS 10.4. The ensemble model showed strong performance, with an AUC ranging from 0.96 to 0.98 and a TSS of 0.84-0.92. Individually, MaxEnt outperformed with an AUC of 98% and a TSS of 92%. The mean temperature of the driest quarter (Bio 9) emerged as the most influential, followed by soil pH and slope. With the current climate conditions, 89.81% of habitats are classified as unsuitable, while 5.64% are least suitable, 2.46% moderately suitable, and 2.09% highly suitable. However, all future projections revealed a decline in suitable habitats, increasing the risk of local extinction. Therefore, it is essential to develop a conservation plan and strengthen climate change adaptation strategies to mitigate habitat loss for this iconic highland species.