SAInvasion: A machine learning model for Spartina alterniflora invasion with high accuracy based on soil macroecological patterns
- Spartina alterniflora invasion increasingly poses a significant threat to the mangrove wetland systems. Preventive control methods have been employed to mitigate the loss of native mangrove plants, facilitated by developing a refined model for predicting species invasions. However, single independent case studies often fail to yield a general conclusion about the predictability of species invasion in specific sediments. In this study, we analyzed 515 samples from 6 different countries to account for the differences between S. alterniflora and non-S. alterniflora sediments using machine-learning methods. The invasion by S. alterniflora shifted microbial diversity, microbial community networks, and microbial processes. Microbial diversity was consistently greater in S. alterniflora sediments. The microbiomes of S. alterniflora sediments harbored higher abundances of Piscirickettsiaceae, Desulfobulbaceae, and Desulfococcus, whereas, non-S. alterniflora sediments contained more Rhodobacteraceae. The distance decay results indicated the weakest dispersal restriction in the microbial community. Furthermore, neutral model results indicated that stochastic processes had the greatest impact on the microbial community in bare beach samples, and the S. alterniflora invasion did not alter the immigration of the microbial community structure. Additionally, 70 bacterial ASVs that classified the species invasion status of the sediment with >80% accuracy were identified using the random forest method. These models can be applied to predict the likelihood of S. alterniflora invasion by revealing key biological indicators and characteristics of the sediment microbiome.
# Method 1. Web pages.
Visit the SAInvasion model website by Clicking https://microbiosee.gxu.edu.cn/ma_sa_predict/.
# Method 2. Raw code.
git clone https://github.com/jinhuili-lab/SAInvasion
cd SAInvasion # Then you can use the model by Rstudio(Posit).
Shuming, Mo, et al. SAInvasion: A machine learning model for Spartina alterniflora invasion with high accuracy based on soil macroecological patterns.