Volume 11, Issue 22 (11-2023)
Ecol Iran For 2023, 11(22): 62-72 |
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:
Bakhtiari S, Rostani Shahraji T, Akhavan R, Ebrahimi Atani R. (2023). Spatial Distribution Modeling of Pistacia atlantica using Artificial Neural Network in Khohir National Park. Ecol Iran For. 11(22), 62-72. doi:10.61186/ifej.11.22.57
URL: http://ifej.sanru.ac.ir/article-1-515-en.html
URL: http://ifej.sanru.ac.ir/article-1-515-en.html
1- Guilan University, Rasht, Iran
2- Research Institute of Forest and Rangelands, Agricultural Research Education and Extension Organization (AREEO), Tehran, Iran
2- Research Institute of Forest and Rangelands, Agricultural Research Education and Extension Organization (AREEO), Tehran, Iran
Abstract: (1312 Views)
Extended Abstract
Background: Investigating the presence of species in forest habitats is crucial for identifying areas suitable for planting and successfully restoring species, as well as understanding the relationship between species presence and environmental factors. This research focuses on predicting the probability of the presence and absence of the Pistacia atlantica forest species in relation to environmental variables, specifically topography and soil science data. The study was conducted in a section of Khojir National Park in Tehran province, covering an area of 120 hectares. Modeling was performed using artificial neural networks and geostatistics to achieve reliable predictions.
Methods: o prepare the necessary maps, slope, aspect, and altitude data were derived from the Digital Elevation Model (DEM) of the study area. Tree sampling in the region was conducted using a regular-random sampling method based on a grid of 100 x 150 meters, resulting in 61 sampling points. Soil sampling was carried out in 17 sample plots, taking into account the diversity of soil conditions by recording the spatial coordinates of each plot. A variety of soil variables were measured in the laboratory, including apparent specific gravity, true specific gravity, absorbable potassium, nitrogen percentage, absorbable phosphorus, organic carbon percentage, electrical conductivity, acidity, soil saturation percentage, lime percentage, and the percentages of sand, silt, and clay. The environmental factors related to soil variables were mapped using geostatistics and GS+ software. Subsequently, a multi-layer perceptron artificial neural network model was designed and validated, correlating the environmental features as model inputs with the presence and absence of Pistacia atlantica as the model output. This validation was conducted using SPSS Modeler software. Finally, based on the model results and the digital map of environmental factors, a prediction map indicating the probability of the presence and absence of Pistacia atlantica was generated using ArcGIS software.
Results: The results of the study revealed that the artificial neural network demonstrated high accuracy, achieving a prediction accuracy of 91% for the presence and absence of Pistacia atlantica. The analysis indicated significant relationships between the presence of Pistacia atlantica and several environmental variables, including electrical conductivity, apparent specific gravity, geographical direction, nitrogen percentage, and altitude. The importance coefficients for these variables were determined to be 0.43, 0.21, 0.17, 0.15, and 0.05, respectively. Furthermore, the agreement between the prediction map and the ground reality map was assessed as good, with a Kappa coefficient of 0.651, indicating a reliable model performance.
Conclusion: In conclusion, the results of this study demonstrate that it is possible to utilize a combination of topographic and soil data to accurately estimate the characteristics influencing the presence of Pistacia atlantica in the researched forests. The generated maps can serve as valuable tools for identifying areas that are conducive to the restoration of this species' habitat. This approach not only enhances our understanding of species distribution in relation to environmental factors but also provides practical applications for conservation efforts aimed at restoring and managing forest ecosystems. By integrating advanced modeling techniques with ecological data, we can better inform strategies for biodiversity conservation and habitat restoration in forested areas.
Methods: o prepare the necessary maps, slope, aspect, and altitude data were derived from the Digital Elevation Model (DEM) of the study area. Tree sampling in the region was conducted using a regular-random sampling method based on a grid of 100 x 150 meters, resulting in 61 sampling points. Soil sampling was carried out in 17 sample plots, taking into account the diversity of soil conditions by recording the spatial coordinates of each plot. A variety of soil variables were measured in the laboratory, including apparent specific gravity, true specific gravity, absorbable potassium, nitrogen percentage, absorbable phosphorus, organic carbon percentage, electrical conductivity, acidity, soil saturation percentage, lime percentage, and the percentages of sand, silt, and clay. The environmental factors related to soil variables were mapped using geostatistics and GS+ software. Subsequently, a multi-layer perceptron artificial neural network model was designed and validated, correlating the environmental features as model inputs with the presence and absence of Pistacia atlantica as the model output. This validation was conducted using SPSS Modeler software. Finally, based on the model results and the digital map of environmental factors, a prediction map indicating the probability of the presence and absence of Pistacia atlantica was generated using ArcGIS software.
Results: The results of the study revealed that the artificial neural network demonstrated high accuracy, achieving a prediction accuracy of 91% for the presence and absence of Pistacia atlantica. The analysis indicated significant relationships between the presence of Pistacia atlantica and several environmental variables, including electrical conductivity, apparent specific gravity, geographical direction, nitrogen percentage, and altitude. The importance coefficients for these variables were determined to be 0.43, 0.21, 0.17, 0.15, and 0.05, respectively. Furthermore, the agreement between the prediction map and the ground reality map was assessed as good, with a Kappa coefficient of 0.651, indicating a reliable model performance.
Conclusion: In conclusion, the results of this study demonstrate that it is possible to utilize a combination of topographic and soil data to accurately estimate the characteristics influencing the presence of Pistacia atlantica in the researched forests. The generated maps can serve as valuable tools for identifying areas that are conducive to the restoration of this species' habitat. This approach not only enhances our understanding of species distribution in relation to environmental factors but also provides practical applications for conservation efforts aimed at restoring and managing forest ecosystems. By integrating advanced modeling techniques with ecological data, we can better inform strategies for biodiversity conservation and habitat restoration in forested areas.
Send email to the article author
| Rights and permissions | |
 |
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License. |
