Distribution Modeling of Hawthorn (Crataegus azarolus L.) in Chaharmahal & Bakhtiari Province Using the Maximum Entropy Method

Introduction
Awareness of the distribution of plant species and their influencing factors has an important role in management, sustainable use and conservation. In particular, habitats and population of species due to increased human degradation, Climate changes and pests and diseases are limited. However, because of the time and budget constraints available to study, there is not enough information available on the distribution of species. Therefore, species distribution modeling (SDM) techniques are an appropriate tool for overcoming these constraints. In these methods, prediction of species distribution from spatial distribution of environmental variables controlling this distribution is possible. Today, using these powerful statistical methods and the GIS, these models are rapidly developing in the field of ecology.
Although a species distribution is influenced by factors such as its ecological niche, its movement ability and inter-species competition, species distribution models focus on environmental factors and ignore the effects of such ecological processes. So far, various methods and models for modeling the distribution of species have been introduced. Most of these methods depend on the presence and absence of species and habitat variables that are related to the ecological niches. One of the strongest and well known models in the distribution of species is the maximum entropy.
Many studies have been carried out using SDMs for animal species by maximum entropy in Iran, but there are fewer studies on plant species and in the case of hawthorn species (Crataegus azarolus L.), any habitat modeling has been done especially with this method as authors investigated. The aim of this study was to fill the gap of the above studies in order to identifying the ecological parameters affecting the spatial distribution of hawthorn species in order to prioritize and provide a map of conservation areas as well as to study the possibility of planting this species in similar areas with the actual site in Chaharmahal & Bakhtiari and better management of existing habitats using maximum entropy analysis method.

Material and methods
Study area
Chaharmahal & Bakhtiari province with an area of 16532 square kilometers located in southwestern of Iran as parts of Zagros Mountain. The absolute maximum temperature is 47.5 °C in Lordegan and the minimum absolute temperature of -34.5 ° C is recorded in Dezak station. The highest precipitation is belonging to the northern highlands with 1600 mm in a year. However, the minimum precipitation is recorded in northeastern parts of the province with an average annual rainfall of 250 to 300 mm, and the average annual precipitation in the province is about 560 mm.

Methods
Species presence data
The criterion for the selection of species occurrence points was the presence of forest stands in which the species had a high density. The initial results from forest survey showed that a large part of the forest areas of the province have Crataegus genus in their composition. Finally, 37 species points were recorded using the Global Positioning System (GPS).

Environmental parameters (variables)
The environmental variables used for the model include 7 climatic variables, 3 topographic variables, and land use and soil variables as in Table 1. The meteorological data of 9 synoptic stations with a period of 15 years (1995-1995) were collected and used. Topographic variables were extracted from digital elevation model (DEM) of the study area with a resolution of 30 m, as well as land use map and soil texture map in a scale of 150,000. Spatial resolution of all layers was changed to 30 m for use in the model and further analysis. After preparing the layers, Maxent 3.3.3e was used for modeling.

Results
The area under the curve (AUC) of the receiving operation factor (ROC) derived from the maximum entropy model was 95%, indicating an excellent prediction of the model versus 5% in the sense of random prediction. The most important variables that have the highest contribution in the model development were height, mean relative humidity and average annual rainfall, respectively.
Jackknife test also showed that the most important variables that have the most contribution in predicting the model are the maximum annual temperature, height, mean annual temperature and mean relative humidity, respectively. Overall, according to the final map, the distribution of Hawthorn species is the best predicted site for the presence of this species located in Ardal, Koohrang and Kayar towns.

Conclusion
The Maxent has proven to be very effective in predicting habitat quantification and distribution of species, since it relies only on species presence data and lacks many of the complications associated with the presence-absence analytical methods. The results obtained from the evaluation of the performance of the Maxent in current study through the AUC of about 95% indicate that the model has excellent predictive capability. The results of the current study showed that even whit the small number of samples (occurrence points), the predictive function of the Maxent can compete with methods that have the highest predictive accuracy and provide acceptable results.
In the process of modeling, it is important to know which variables and to what extent they have played a role in modeling. The results of species performance along the slope of environmental changes through the response curves are obtained and based on the analysis that in relation to the factors influencing the distribution of hawthorn species in this study, height, mean relative humidity and average annual sum of rainfall were the most important factors influencing the distribution of hawthorn species in Chaharmahal & Bakhtiari province.
According to the response curve of hawthorn in relation to height, the most probable presence of species in areas with a height of 1000 to 2000 meters is predicted, so that with the increasing of height the likelihood of presence of the species is reduced, which is consistent with the results of the research in the forests of Abdanan, Ilam Province. The response curve of hawthorn in relation to soil type also showed that the most probable presence of hawthorn is in the loam and silty soils. The highest presence of hawthorn species based on the response curve in the slope is about 0-2% and with increasing slope, the probability of prediction of species distribution decreases. Also, the results of hawthorn response curve to the land use indicated that the probability of species presence in areas with arable land, pasture and forest use is more than other uses.
Finally, the results of this research provided important information about the range of tolerance of Hawthorn species to the influential environmental variables. This information is effective in making management decisions to prioritize conservation areas and to improve conservational measures, particularly in areas where vegetation is degrading, and increases the chance of success in planting and restoration projects.