Tecomella undulata under threat: The impact of climate change on the distribution of a valuable tree species using a machine learning model

PLOS One | https://doi.org/10.1371/journal.pone.0326609 July 9, 2025 7 / 18 Results Importance of variables and model performance The MaxEnt model achieved excellent predictive performance (Fig 2) for the distribution of the T. undulata species, indicating that it can accurately predict the distribution of this species. The TSS value of 0.79 further confirms the mod- el's good predictive accuracy. The distribution model for T. undulata demonstrates that varied bioclimatic and geographical factors contribute substan- tially to its potential distribution (Fig 3). Temperature seasonality (bio4) holds prominent importance, contributing to 35.9% of the potential distribution of T. undulata. This points towards the species’ high sensitivity towards seasonal temperature fluctuations. Mean temperature of the wettest quarter (bio8) comes in second, accounting for 23.2% of the distribution. This shows the species’ affinity for warmer climates during the wettest quarters of the year. Precipitation seasonality (bio15) contributes 12.2%, signifying T. undulata’s dependence on precipitation fluctuations throughout the year. Mean temperature of the driest quarter (bio9) provides 11.3% of potential distribution insights, highlighting the species’ adaptabil- ity to temperature ranges in such periods. Precipitation of the driest (bio14) and coldest quarters (bio19) contribute 10.2% and 10.1% respectively, suggesting the species’ resilience during dry and cold periods. The digital elevation model (DEM)

also serves as a significant factor with a contribution of 8.9%. This notably implies the species’ predisposition towards cer- tain elevations. Mean diurnal range (bio2) plays a more moderate role, comprising 5.8% of the distribution. This demon- strates how daily temperature ranges can influence the species’ distribution. Finally, isothermality (bio3), contributing a marginal 3.8%, indicates the least effect on the species’ distribution. From the contribution rates, we can observe that temperature factors including bio4, bio8, bio9, bio3, and bio2 col- lectively contribute a significant portion (80%) towards the potential distribution of T. undulata, underlining the species’ sensitivity towards temperature. On the other hand, the collective weight of precipitation factors such as bio15, bio14, and bio19 amount to 32.5%, proffering moisture’s role in the species’ geographic spread. Alongside bioclimatic influences, the species shows a marked preference for certain altitudes as per the Digital Elevation Model (DEM). These findings high- light the complex interplay of varied ecological factors in shaping the distribution of T. undulata. Fig 2. Area under the receiver operating characteristic curve (AUC) for MaxEnt model. https://doi.org/10.1371/journal.pone.0326609.g002