Predicting the impacts of climate change on Persian oak (Quercus brantii) using Species Distribution Modelling in Central Zagros for conservation planning

Document Type : Research Paper


1 Faculty of Natural Resources, Sari Agricultural Sciences and Natural Resources University, Sari, I.R. Iran.

2 Assistant Prof., Faculty of Natural Resources, Faculty of Natural Resources, Sari Agricultural Sciences and Natural Resources University, Sari, I.R. Iran.

3 Associate Prof., Faculty of Natural Resources, Sari Agricultural Sciences and Natural Resources University, Sari, I.R. Iran.

4 Assistant Prof., Faculty of Natural Resources, Isfahan University of Technology, Isfahan, I.R. Iran.


In recent years, climate change has affected on both ecosystems and the creatures that live in them. Hence, plant species may be expected to show marked redistributions in reply to climate change; this has estimated at various scales and in diverse places, usually by the use of bioclimatic envelope models. These models are often named distribution models (SDMs), Climate envelope models or ecological niche models (ENM). They use climate factors as independent predictor variables and biotic data as dependent variables to produce a predictive model for species or ecosystem distributions. Climate envelope models can be informative, not by forecasting changes, but by quantifying differences among current distributions and potential habitats under potential climate change scenarios. Moreover, despite the deficiencies of climate envelope models, the overall patterns of predicted species range shifts often match observed biological tendencies. Temperature over south-west Iran may increase between 1.69 and 6.88 °C by 2100. Summer temperatures may increase with higher rates than spring, winter, and autumn temperatures. The main genus in the Zagros region is oak with a varied range of species distributed across the area. Quercus brantii Lindl. Species is known as Persian oak. Persian oak species is endemic to temperate regions of Asia and western Asia, including Iran, Iraq, Syria, and Turkey and it’s the boundary of Irano- Turanian vegetation region. Persian Oak is a prevalent species in Zagros forests. In the Zagros showed that forest loss was more closely associated with climate change and urban human population increase. Recently, a large number of deaths among oak trees have been reported in their natural habitats. Therefore, this study aimed at predicting the effect of climate change on the geographical distribution of Persian Oak in Chaharmahal & Bakhtiari province in the central Zagros region under future climate scenarios by 2050.

Materials and methods:
The studied region in this research had the area of 1.6 million hectares in the central Zagros in Chaharmahal & Bakhtiari province. Field studies, including harvest of geographical coordinates of the presence of this species in Central Zagros. In this study, 19 bioclimatic variable climate factors describing Persian Oak habitat, which have been used in many studies as the basis for monitoring impacts of climate change on organisms. Climate change RCP4.5 scenario, general circulation model HadGEM2-CC and three physiographical variables (elevation, slope and slope aspect) were applied to. These 19 bioclimatic layers must be derived from the three basic climatic variables. WorldClim provides monthly maximum (Tmax), minimum (Tmin), and mean temperatures, and monthly precipitation. Monthly precipitation was improved by average monthly precipitation obtained from weather stations across the province. Then 19 bioclimatic variables were created in DIVA-GIS. Digital Terrain Model (DTM) was used to generate slope and aspect data layers and was used as physiographic variables. First, the presence of the correlation between variables modified by Pearson test was examined and the variables with over 80% correlation with each other were determined. After conducting Pearson's correlation analysis and removing variables with high correlation, it was found that 8 variables (BIO9, BIO7, BIO3, BIO4, BIO12, and BIO17, percent of slope and slope aspect) were not correlated with each other and can enter the final model. In this regard, we used 5 modelling approaches, Generalized Linear Model (GLM), Classification Tree Analysis (CTA), and Artificial Neural Network (ANN), Generalized Boosting Method (GBM) and Random Forest (RF) to determine relationships between the occurrence of species and environmental factors. One difficulty with the use of species distribution models is that the number of techniques available is large and is increasing steadily, making it difficult for the user to select the most appropriate methodology for their needs. This is particularly true when models are used to project distributions of species into independent situations, which is the example of projections of species distributions under future climate change scenarios. In this study, we used the ensemble predictions of the models under the framework Biomod and R software. To do so, 20% of the species presence data was devoted to the evaluation and 80% to the implementation of these models. This process was repeated for 10 times for each of the used models; at each time, evaluation and implementation parts were randomly selected from the data. Finally, for each model, the results were obtained from these 10 runs. The predictive model performance was evaluated using one main kind of accuracy measures, Area under the receiver operating characteristic curve.

Results and discussion:
All models assume a certain equilibrium, namely that the species occurs in all environments where it is possible to survive, that it cannot survive outside this range and that it is in equilibrium with climate. In fact, due to many reasons (time delay in response, limited dispersal, anthropogenic influence), the situation is probably different from many species. Therefore, we report here the potential changes in suitable habitat for Persian oak, not the real range changes that will happen. However, a certain trend can be observed, which should influence decisions to be prepared for the species’ response to climate change. Monitoring using Species distribution modelling can help us distinguish the most important factors in determining species presence and in designing conservation programs. This research showed Annual precipitation (49.7%) and mean temperature of the driest quarter (27.7%) have played the most important role in habitat suitability of this species. Under RCP4.5 climate scenario Persian Oak might lose 35.7% of its climatically suitable habitats due to climate change factors, by 2050, while in a number of areas (61.4%), the currently unsuitable habitats may be converted into suitable. In the studied region as a result of climate changes Persian Oak was moved to the higher elevation, Results have been similarly obtained in many of the studies in which the movement of species affected by climate change has been studied. Among all the statistical techniques, RF was found to be the most reliable model for species prediction. However, the predictions from the different models varied a lot, even if for one given species, outcomes of prediction may change from model to model. In an opinion that each predictive model relied on different mathematical functions, SDM will give a variety of results without the doubt. Nevertheless, it was not so surprising because the RF model gives the predictions by producing thousands of trees and aggregated with an average. Thus, in this research RF was a robust technical modelling for species distribution prediction and ensemble modelling was also regarded as the best solutions to reduce the single model uncertainties and bias. To assess the accuracy of the maps produced by the model, AUC of ROC plot was used. Based on these results, all models are functioning well, because whenever an implemented model has the AUC values of more than 0.7 and 0.9, it will be considered a good and excellent model, respectively; otherwise, the model is weak.
This research showed that ensemble modelling by Biomod could predict the current potential distribution of Persian Oak with high accuracy (AUC varies from 0.85 to 0.98). Among all the statistical techniques, RF was found to be the most reliable model for species prediction. This research showed Annual precipitation and mean temperature of the driest quarter have played the most important role in habitat suitability of this species These models could also predict the geographical shift of given species under climate change scenarios (RCP4.5). According to the results, Persian Oak is expected to move toward higher elevation, with a decreasing area of the current distribution. This study highlights the importance of climate change on the geographical plant species distribution. Persian Oak is one of the most important forest plant species in the central Zagros of Iran, very crucial for soil conservation and biodiversity; hence, it requires an extra effort to protect species such as Persian Oak against climate change.


Main Subjects