TL;DR
This study uses the ensemble modeling approach Biomod2 to predict the current and future suitable habitats for three endangered Fritillaria species in China under climate change. The findings highlight temperature seasonality and isothermality as key environmental drivers and reveal a significant correlation between species' spatial distribution and their chemical composition, offering crucial insights for conservation and cultivation strategies.
What Problem Does This Paper Address?
Fritillaria species, particularly those used in traditional Chinese medicine (TCM) as "Chuan-Bei-mu," are facing severe population declines due to overharvesting. This scarcity poses a significant threat to both their medicinal supply and biodiversity. To address this, understanding their current ecological distribution and predicting how these habitats will change under future climate scenarios is critical for developing effective conservation and cultivation strategies. Previous studies have often relied on single modeling approaches, which can limit prediction accuracy. This paper aims to overcome this by employing an ensemble modeling technique and investigating the interplay between environmental factors, species distribution, and their chemical properties.
Methods
The researchers employed the Biomod2 package in R as their primary species distribution modeling (SDM) tool. Biomod2 is an ensemble modeling framework that integrates multiple algorithms (including MaxEnt, Random Forest, and others) to improve prediction accuracy and robustness. They compared Biomod2's performance against individual MaxEnt and GARP models using standard evaluation metrics like AUC, KAPPA, and TSS.
Key methodological steps included:
- Occurrence Data Collection: Gathering records from field surveys, literature, and online databases.
- Environmental Variable Selection: Utilizing 19 bioclimatic variables from WorldClim and three topographic variables (slope, altitude, aspect). Spearman correlation was used to select the most informative and non-redundant variables.
- Model Training and Evaluation: Training the models with various data splits (85/15, 80/20, 75/25) and selecting the optimal model based on high AUC, KAPPA, and TSS values.
- Future Climate Projections: Using four Shared Socioeconomic Pathways (SSPs) from CMIP6 (SSP126, SSP245, SSP370, SSP585) to project habitat changes from 2021 to 2100.
- Niche Overlap Analysis: Quantifying the overlap in ecological niches among the three species using Levins' B1, B2, Schoener's D, and Hellinger's I indices.
- Chemical Composition Analysis: Performing clustering analysis based on chemical metabolites from literature and databases.
- Correlation Analysis: Investigating the relationships between DNA phylogenies, chemical composition clustering, and niche overlap using linear trend lines and Pearson correlation tests.
Datasets
- Occurrence Records: 118 individual sample points for Fritillaria delavayi (56 points), Fritillaria taipaiensis (45 points), and Fritillaria wabuensis (17 points) collected from field surveys, literature, and online databases (Global Biodiversity Information Facility - GBIF, Chinese Virtual Herbarium - CVH, Plant Science of China). Records spanned from 1884 to 2020.
- Environmental Variables:
- 19 bioclimatic variables from WorldClim database (version 2.0) at 2.5 arc-min spatial resolution.
- 3 topographic variables (slope, altitude, aspect) extracted using ArcGIS.
- Future Climate Data: Projections from CMIP6 under four Shared Socioeconomic Pathways (SSPs): SSP126, SSP245, SSP370, and SSP585, at 2.5 arc-min resolution for the period 2021-2100.
- Chemical Metabolite Data: Obtained from the Traditional Chinese Medicine Integrated Database (TCMID) and academic literature.
Key Insights and Ideas
- Ensemble Modeling Superiority: Biomod2 consistently outperformed single models like MaxEnt and GARP, demonstrating the value of ensemble approaches for robust ecological niche modeling.
- Key Environmental Drivers: Temperature seasonality (Bio3 for F. delavayi) and isothermality (Bio4 for F. taipaiensis and F. wabuensis) were identified as the most influential factors for species distribution, underscoring the sensitivity of these plants to temperature fluctuations.
- Climate Change Impacts: Future projections indicate shifts in suitable habitats, with some species benefiting from warmer conditions under certain scenarios (e.g., F. taipaiensis under SSP126), while others might face challenges (e.g., F. delavayi under SSP585).
- Niche Overlap and Chemical Correlation: A strong positive correlation was found between the ecological niche overlap of the Fritillaria species and their chemical composition. This suggests that plants sharing similar habitats and environmental conditions tend to produce similar secondary metabolites.
- Distribution Influences Chemistry: Conversely, no significant correlation was found between niche overlap and DNA barcodes, implying that spatial distribution and environmental factors have a more direct influence on chemical composition than genetic relatedness alone. This has implications for understanding the variability of medicinal compounds.
- Conservation Implications: Identifying species-specific habitats is crucial for reducing inter-species competition and developing targeted conservation and cultivation strategies.
Key Results
- Model Performance: Biomod2 achieved excellent performance across all three species, with AUC values generally above 0.99, KAPPA above 0.95, and TSS above 0.96.
- Current Suitable Habitats:
- F. delavayi: 17.72% of China's total area.
- F. taipaiensis: 23.06% of China's total area.
- F. wabuensis: 20.60% of China's total area.
- Future Habitat Projections:
- F. delavayi and F. wabuensis showed maximum suitable habitat under the SSP585 scenario.
- F. taipaiensis showed maximum suitable habitat under the SSP126 scenario.
- Niche Overlap: High niche overlap was observed among the three species, with F. wabuensis and F. delavayi showing the highest overlap.
- Correlation: A strong correlation (R² = 0.994, P=0.050) was found between niche overlap and chemical composition.
Limitations and Future Work
- Data Limitations: The study acknowledges that economic or other environmental factors (like soil salinity) were not considered, which could influence species distribution and growth. The limited number of occurrence records for F. wabuensis might also impact model precision.
- Future Work: The authors suggest that incorporating more comprehensive environmental data and exploring the influence of factors like soil microbial communities and land-use changes would enhance prediction accuracy. Further systematic studies on chemical metabolite profiles, especially for less-studied species, are needed to refine the understanding of the spatial distribution-chemical composition link. The study also points to the need for more detailed investigations into the evolutionary drivers of species diversification, as spatial distribution and phylogeny showed weak correlation. The identification of species-specific habitats and future distribution patterns provides a foundation for developing sustainable conservation and cultivation strategies for these valuable medicinal plants.
Summary updated