Volume 17, Issue 4 (2-2024)
مرتع 2024, 17(4): 513-528 |
Back to browse issues page
Download citation:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
BibTeX | RIS | EndNote | Medlars | ProCite | Reference Manager | RefWorks
Send citation to:
Jafarian Z, amiri M. Investigating the Potential Habitat of Bromus stenostachyus Boiss. in Mazandaran Rangelands Using an Ensemble Modeling Approach. مرتع 2024; 17 (4) :513-528
URL: http://rangelandsrm.ir/article-1-1159-en.html
URL: http://rangelandsrm.ir/article-1-1159-en.html
Department of Rangeland Management, Faculty of Natural Resources, Sari Agricultural Sciences and Natural Resources University, Sari
Abstract: (398 Views)
Background and objectives: Environmental factors play a pivotal role in determining the flora and fauna distribution in a region, contributing to increased ecosystem biodiversity and species composition on a large scale. Given the intricate nature of natural systems, species distribution models are employed to comprehend the impact of these factors on species' potential niches. In recent studies, bioclimatic variables from the CHELSA database (1979-2013), derived from temperature and precipitation data interpolation, have been widely used for this purpose.
Methodology: This study investigates the potential distribution and preference tendencies of Bromus stenostachyus in the Hyrcanian rangelands of Mazandaran province using CHELSA bioclimatic variables, along with physiographic and anthropogenic factors. Stratified random sampling, based on physiognomy-florestic structure, was employed to collect species occurrence data at one square kilometer resolution. After selecting key environmental variables, an ensemble method incorporating seven modeling algorithms—Generalized Linear Model (GLM), Gradient Boosting Machine (GBM), Multivariate Adaptive Regression Spline (MARS), Random Forest (RF), Artificial Neural Network (ANN), Maximum Entropy (Maxent), and Classification Tree Analysis (CTA)—was utilized. Model performances were evaluated using area under the curve (AUC) and threshold-dependent measures such as True Skill Statistic (TSS), Sensitivity, and Specificity. Sensitivity analysis was conducted to determine the importance of environmental variables, and species response curves were plotted based on the most efficient model to identify optimal ecological conditions.
Results: Based on validation criteria, all the fitted models showed excellent performance to predict the species habitat. The ensemble model outperformed single models, and among single models, GBM and RF showed the best performance. According to the results of sensitivity analysis, maximum temperature of warmest month (Bio5) was determined the most important variable to fit the species distribution models, in a way that explained more than a quarter of changes in the distribution. After averaging ten model runs, the species potential distribution map was produced under single models and the ensemble model. Because the prediction of each model depends on its mathematical functions, the distribution models had different results. But in general, the species habitats are distributed in parts of province where the elevation and precipitation are higher than the average elevation and precipitation in the whole province, but their temperature is lower than the average temperature of the province. On the response curves, the optimum habitats for the species were located in areas where the maximum temperature of warmest month is less than 23 °C, temperature annual range is less than 32 °C, precipitation of wettest month is at most 100 mm, isothermality is less than 26 and temperature seasonality is in the range of 63-68.
Conclusion: The ensemble modeling approach yields more realistic simulations of species distribution by emphasizing common areas of agreement among single models. The study results can inform practical planning for the preservation and management of Hyrcanian rangelands, as well as suggest suitable species for reclamation, rehabilitation, and restoration efforts in ecologically similar areas. These findings facilitate optimal exploitation and informed management decisions while preserving native species.
Methodology: This study investigates the potential distribution and preference tendencies of Bromus stenostachyus in the Hyrcanian rangelands of Mazandaran province using CHELSA bioclimatic variables, along with physiographic and anthropogenic factors. Stratified random sampling, based on physiognomy-florestic structure, was employed to collect species occurrence data at one square kilometer resolution. After selecting key environmental variables, an ensemble method incorporating seven modeling algorithms—Generalized Linear Model (GLM), Gradient Boosting Machine (GBM), Multivariate Adaptive Regression Spline (MARS), Random Forest (RF), Artificial Neural Network (ANN), Maximum Entropy (Maxent), and Classification Tree Analysis (CTA)—was utilized. Model performances were evaluated using area under the curve (AUC) and threshold-dependent measures such as True Skill Statistic (TSS), Sensitivity, and Specificity. Sensitivity analysis was conducted to determine the importance of environmental variables, and species response curves were plotted based on the most efficient model to identify optimal ecological conditions.
Results: Based on validation criteria, all the fitted models showed excellent performance to predict the species habitat. The ensemble model outperformed single models, and among single models, GBM and RF showed the best performance. According to the results of sensitivity analysis, maximum temperature of warmest month (Bio5) was determined the most important variable to fit the species distribution models, in a way that explained more than a quarter of changes in the distribution. After averaging ten model runs, the species potential distribution map was produced under single models and the ensemble model. Because the prediction of each model depends on its mathematical functions, the distribution models had different results. But in general, the species habitats are distributed in parts of province where the elevation and precipitation are higher than the average elevation and precipitation in the whole province, but their temperature is lower than the average temperature of the province. On the response curves, the optimum habitats for the species were located in areas where the maximum temperature of warmest month is less than 23 °C, temperature annual range is less than 32 °C, precipitation of wettest month is at most 100 mm, isothermality is less than 26 and temperature seasonality is in the range of 63-68.
Conclusion: The ensemble modeling approach yields more realistic simulations of species distribution by emphasizing common areas of agreement among single models. The study results can inform practical planning for the preservation and management of Hyrcanian rangelands, as well as suggest suitable species for reclamation, rehabilitation, and restoration efforts in ecologically similar areas. These findings facilitate optimal exploitation and informed management decisions while preserving native species.
Type of Study: Applicable |
Subject:
Special
Received: 2022/08/16 | Accepted: 2023/04/16 | Published: 2024/02/29
Received: 2022/08/16 | Accepted: 2023/04/16 | Published: 2024/02/29
Send email to the article author
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |
