Model performance
Ensemble models were widely adopted in many researches related to
species distribution(Hao, Elith, Lahoz-Monfort, & Guillera-Arroita,
2020). Previous studies showed that uncertainty of model result would be
brought out by many factors such as species distribution models,
dispersal strategies, environmental factors and so on(W. Thuiller,
Guéguen, Renaud, Karger, & Zimmermann, 2019). Although there were lack
of strong quantitative conclusions about the predictive performance of
ensemble models(Hao, Elith, Guillera-Arroita, & Lahoz-Monfort, 2019).
There was a popular belief that ensemble model would reduce the
uncertainty due to sole algorithm, and it would perform better than sole
one(Araújo et al., 2019; Araújo & New, 2007). In our study, except SRE,
the AUC values of all the other nine models were greater than 0.8, and
the TSS values were greater than 0.6. The values of ensemble model were
0.971 and 0.835 respectively. Based on this, an obvious result could
easily draw out that ensemble model outperformed any sole algorithm,
which was consistent with the conclusion ahead. This result also showed
that the simulation was with high accuracy and could be used to predict
the current and future potential distributions of A. annua .