Predicting suitable habitat of Egyptian Vulture (Neophron percnopterus) in Iraq, using Maxent model

Master's Thesis, 2014
53 Pages, Grade: 53.60


Table of Content


Chapter 1: Introduction

Chapter 2: Methods
2.1.The Study area:
2.2.The Egyptian Vulture:
2.2.1 Ecology and Habitat Description of Egyptian Vulture
2.3.Occurrence data:
2.4.Environmental data
2.5.Data Analysis
2.5.1 Developing the model
2.5.2 Maxent
2.5.3 Maxent output data format
2.5.4 Model evaluation

Chapter 3: Result
3.1.Response Curves results:
3.2. Jackknife results:

Chapter 4: Discussion
4.1. Response Curve and Environmental Variables contribution
4.3.Conservation Implication



List of Figures, and Tables


1. Distribution of Egyptian Vulture across the world

2. Observed records of Egyptian Vulture

3. Receiver operating characteristic curve (ROC)

4. Result from Maxent, right figure before reclassified of predicted suitable, left figure after reclassified of values with 10 percentile training presence

5. Response curve from Maxent for the most important variables for the species distribution model of Egyptian Vulture

6. Types of land cover in Iraq

7. Gains of the variables in the Maxent model (jackknife test)

8. Key Biodiversity Areas of Nature Iraq survey sites with observed record of Egyptian Vulture in Iraq

9. Unknown suitable Habitat for Egyptian Vulture

10. Know suitable habitat for Egyptian Vulture

11. Compare between the know habitat (in KBAs) and unknown habitat.

12. Precipitation map of Iraq


1. Showed environmental variables

2. Land cover of Iraq (ESA, 2009)

3. Relative contributions of the environmental variables to the Maxent model


Species distribution modeling to predict and map appropriate habitat for endangered species is critical for monitoring and restoration of their population decline in natural habitat. However, data on distribution of endangered species are limited. Egyptian Vulture (Neophron percnopterus) is one of the endangered species in the world that its population has been declining by 50% in the last three decade. In this study, we have tried to test the predicted distribution of potentially suitable habitat for Egyptian Vulture in Iraq by using Maxent as species distribution model. To achieve these aims, we could be done with species distribution model outputs of Egyptian Vulture from occurrence data assembled from Nature Iraq and made distribution predictions, demonstrating how landscape and environmental variables may affect the species distribution and We determined their importance and compared known habitat of Egyptian Vulture from KBAs with predicted habitat predictions fromMaxent. The model performed very well, The AUC value is 0.870, which is better than random prediction. The results indicate that one of the precipitation inputs, bio_13, contributed 54.3% to the final description of potential habitat for the Egyptian Vulture. This provides evidence that precipitation, particularly precipitation of wettest month, is the most significant variable in the model’s determination of suitable habitat locations. Several of the highest suitability points predicted by Maxent are associated with known Egyptian Vulture presence and some of them match with KBAs. But, there are a number of locations, which were predicted to have high suitability of habitat for Egyptian Vulture, are located outside KBAs. This study is important to do more fieldwork for predicted high suitable area for Egyptian Vulture in the future and protect them from population decline.

Chapter 1: Introduction

In recent years, species distribution has become a significant constituent in the planning of conservation, for this purpose many different technical models have been established(Guisan & Thuiller, 2005; Pearson, 2010). These models can be used to recognize the environmental state of the species population between environmental variables within their presence record of that species (Pearson, 2010). Species distribution modeling is a statistical modeling approaches to predict the spatial distribution of definite species, and has important recognition as a part of conservation planning currently (Araújo, Cabeza, Thuiller, Hannah, & Williams, 2004; Austin, 2002; Braunisch, Patthey, & Arlettaz, 2011; Gastón & García-Viñas, 2011; Kremen et al., 2008; Tarjuelo, Morales, Traba, & Delgado, 2014). For example, distribution modeling could be applied to predict spatial distribution modification of the species during of habitat fragmentation and climate changes, persevering significant ecological occurrence that could endanger the species in apprehension. Species distribution modeling permits the operation of widely available distributional data (e.g., occurrence records) to predict distribution. Compared to the traditional landscape ecology field studies, ecological modeling provides the fundamental to study ecological questions over widespread range of spatial and chronological scales within a short time span and minimal resources available(Lindenmayer, Cunningham, & Donnelly, 1994; Starfield & Bleloch, 1986). Application of species distribution modeling to predict and mapping appropriate habitat for endangered species is critical for monitoring and restoration of their decline population in natural habitat. However, data on distribution of endangered species are limited (Kumar & Stohlgren, 2009).

Maxent is a presence-only modeling method with a confirmed attested potential to predict wildlife distribution. This approach achieves very good in predicting spatial distribution of species data and predicting suitable habitat (Pearson, Raxworthy, Nakamura, & Townsend Peterson, 2007; Phillips, Anderson, & Schapire, 2006). Because of, Maxent has some advantages which has better performs than much other modeling methods when comparing with them, and this is due to the numbers benefits of the model for example it needs only presence data with environmental layers, it can be run with very few occurrence data (Hernandez et al., 2008; Pearson, Raxworthy, Nakamura, & Townsend Peterson, 2006), similarly it can be integrate interaction between variables and using continuous or categorical data, it developed effectively to coverage to maximum entropy probability distribution as a guaranteed and it can be avoided over-fitting by using regularization (Ortega-Huerta & Peterson, 2008; Phillips et al., 2006; Redon & Luque, 2010; Wang, Xie, Wan, Xiao, & Dai, 2007; Ward, 2007). However, Maxent has some drawbacks for example it is not a perfect statistical method as GLM or GAM; there are fewer instructions for use and fewer methods for valuing the amount of error in a prediction. For effectiveness in over-fitting variable with other selection methods, it needs to more study in the amount of regularization. Furthermore, Maxent is not a standard statistical package therefore it requires special-purpose software.

Egyptian Vulture (Neophron percnopterus) is one of the endangered species in the world that its population has been declining in the world by 50% in the last three decade(BirdLife, 2014b) .The massive decrease of its population is located in the breeding area (Figure 1). Therefore, it is documented as endangered species in IUCN red list since 2007 (Angelo, Hashim, & Oppwl, 2013; BirdLife, 2014a; I angelov, 2013) .The reasons for this huge decline are due to changing in their habitat, human disturbance, electrocution(Zuberogoitia, Zabala, Martínez, Martínez, & Azkona, 2008), wind-farm(Carrete, Sánchez-Zapata, Benítez, Lobón, & Donázar, 2009), poising by drug (e.g. Diclofenac) and lead (e.g. from gun shot) (BirdLife, 2014a; L. Gangoso et al., 2009; Liberatori & Penteriani, 2001; Milchev & Georgiev, 2014). But, in some places in the world for example in Socotra their population has been stable or increases because of there is no disruption for vultures, no indication of direct poisoning, and no obstinate pesticides are used in farming applies (I angelov, 2013; Porter & Suleyman, 2012).

In Iraq, especially in Kurdistan Region; Egyptian Vulture was listed as threatened species since 1994 by Evan, in Benavi, Dori Serguza, and Ser Amadia (Dohuk governorate); Bakhma, Dukan and Darbandikhan Dams (Dohuk, Erbil, Sulaimani); and in Mahzam and Lake Tharthar (Salah Al Din), (Evans, 1994). Nature Iraq organization have been collected data on birds throughout the country in order to recognize sites that are globally, regionally, and/or nationally important for their biological diversity and support for their protection.

In this study, we have tested the prediction distribution of potential suitable habitat for Egyptian Vulture in Iraq by using Maxent as species distribution model, which is a new method developed by Phillips (2006). Therefore, to conserve this endangered species it needs extensive fieldwork, which is inaccessible and economically invaluable, because Iraq has divided into four different regions: the lower Mesopotamia, desert plateau, upper plains and foothills and Mountains (Evans, 1994). Therefore, species distribution modeling is a solution to predict suitable habitat of the species and currently it’s important methods in wildlife management and conservation planning(Calabrese, Certain, Kraan, & Dormann, 2014). Maxent has a place as well, using a case study of Egyptian Vulture distribution in Iraq derived from. Therefore, it can be doing fieldwork to the predicted suitable habitat, which is predicted by Maxent and protected the predicted suitable habitat, which is more valuable economically.

This study aimed to test the prediction distribution of potential suitable habitat for Egyptian Vulture in Iraq by using Maxent as species distribution model. The study explored occurrence data records of Egyptian Vulture to create ecological models. The analysis focused on the Iraq generally and Kurdistan Region particularly. Through the ecological model outputs, determine and investigate the importance of landscape, environmental variables effects to Egyptian Vulture distribution. To achieve these aims, hence, demonstrate how landscape and environmental variables affect the species distribution. Determine their importance and compare known habitats of Egyptian Vulture from KBAs with the result of predicted habitats.

Chapter 2: Methods

2.1.The Study area:

Iraq is located between 29° 5’ and 37° 22’ N latitude and 38° 45’ and 48° 45’ E longitudes in the southwest Asia continent (FAO, 2011; "Maps of World," 2013). It’s characterized by distinctive environmental, biological and social structures, which are different anywhere else in the Arabian Peninsula and it is located in two ecological region which are Mediterranean forests, woodlands and scrubs with Mesopotamian delta and marshes(WWF, 2014). It has 600 miles (965.61 km) in length from the northern border to Persian Gulf, and its width about 450 miles (4267.2 km), (FAO, 2011). The latitude is range from sea level from south to peaks of mountains in the north, which reach 14.000 feet (4267.2 m) as height, and it covers 438 317 km2. It surrounded by Iran from the east, Saudi Arabia and Kuwait from south, Syria and Jordan from the west and Turkey from the north. In addition, Iraq divided into four main biogeographically region; lower Mesopotamia, which is extended from foot of Jabal Hamrin to the northern east of Arabian Gulf, Desert plateau, which is located in the south-west of the Euphrates river, Upper plains and Foothills, which extended through the country in a north-westerly direction from Mansuriya (100 km north-east of Baghdad), the last region is Mountains region, which is located in the north and north-east border of Iraq, and extended from Zakho to Halabja. The climate of Iraq is described by a cool rainy winter with a hot dry summer(Evans, 1994; Hatt, 1959).

2.2.The Egyptian Vulture:

Egyptian Vulture (Neophron percnopterus) is belonging to kingdom of Animalia, phylum Chordate, class Aves, order Falconiformes (Accipitriformes) and family Accipitridae (BirdLife, 2014a)(Linnaeus, 1758). It is medium to large size (Length 55-65 cm, Wingspan 155-170 cm) raptor, which is characterized by flight silhouette with broad and well-fingered wings with a wedge-shaped tail. It has a yellow face with a black tip bill. The plumage of adult is mostly a pale grey to whitish within primaries and secondaries are black in color in addition to the some buff on the head and neck. But the Juveniles are mainly dark brown with pale buff, also it has a V-shape tail when flying (BirdLife, 2014b; Svensson, Grant, Mullarney, Zetterström, & Christie, 2009).

2.2.1 Ecology and Habitat Description of Egyptian Vulture

Egyptian Vultures is a breeding summer visitor to the mountains and rocky hills in northern and western Iraq; also a passage migrant(Salim, Al-Sheikhly, Majeed, & Porter, 2012). Over most of range of Egyptian Vultures typically are nests on ledges or in caves cliffs, crags and rocky outcrops; otherwise survives and feeds mostly in any lowland or montane open area, often-arid country, also it is a territory bird and the breeding pairs defend their active nest with using the same territory each year (J. A. Donázar, Ceballos, & Tella, 1994; Ferguson-Lees & Christie, 2001; Svensson et al., 2009). In addition, it roost on large trees (e.g. in Socotra, Yemen)(Porter & Suleyman, 2012) and North Spain and South France(José A Donázar, Ceballos, & Tella, 1996), nesting on building in India, electricity towers (Naoroji & Schmitt, 2007) and unusually on the ground(Laura Gangoso & Palacios, 2005). Moreover, along beaches, river, sandbanks, wetland edge, and frequently near human habitation: thus desert edge, the African Sahel, high rocky plains and ravines, steppe and other grassland, open savannah (but not closed woodland), cultivation, rubbish dumps, harbours and villages(Ferguson-Lees & Christie, 2001). Also scavenges at human settlements. It is an opportunistic scavenger, which has a wide range of diet including carrion, tortoises, organic waste, insects, young vertebrates, eggs and even feces (Ferguson-Lees & Christie, 2001). It occupies a large range of the world, their distribution extended from India subcontinent to the Europe and Africa continents (BirdLife, 2014a, 2014b)( Figure 1). In Iraq, according to the literature and Nature Iraq surveys between 2007-2010, it mostly distributed in Kurdistan northeast of the country(Allūs, 1960; Evans, 1994; Iraq, 2011; Moudhafer et al., 2006).

2.3.Occurrence data:

The occurrence data points for Egyptian Vulture (Neophron percnopterus) were downloaded in the word bird website, which is collected by Nature Iraq bird team during bird survey in the project called (Key Biodiversity Area) between 2007-2010 and download the KBAs map (KML files) from Nature Iraq website (Iraq, 2014). We used 99 data points and we put in to the Geographical Information System (GIS) version 10.2 ( to made a layer for presence Egyptian Vulture distribution and KBAs map with GIS. In addition, we put the coordinates to Microsoft excel in order to make a Comma Separated Values (CSV) file which is needed by Maxent program in sample file.

2.4.Environmental data

We used 21 environmental coverage variables to model the potential distribution of the Egyptian Vulture (Neophron percnopterus), Twenty bioclimatic variables related to temperature, precipitation and altitude in raster format with 30 -seconds resolution, which is roughly equivalent to 1 km2 cells (Table 1) downloaded from WorldClim–Global Climate Data ( These data are a set of climate layers between (1950-2000 years) that represent information derived from monthly temperature and rainfall obtained from weather stations and then interpolated for illustrating the average value of surfaces with a spatial resolution of 1 km2 (Hijmans & Graham, 2006) (Hijmans, Cameron, Parra, Jones, & Jarvis, 2005) and create more biological significant. Moreover, we used vegetation types as land cover for Iraq, which is downloaded from the website of European Space Agency GlobCover (, version 2.3-2009. WorldClim and Landover variables are global, so we made the necessary cuts to cover only our country Iraq; by GIS-Arc toolbox-Spatial Analyst Tool, the layers were extracted by mask and rescale to our country. Furthermore, they layers were converted from Raster format to ASCII format by conversion tools. The 21 environmental variables were resamples to the WGS84 geographical coordinate system at a resolution of ~1 km2.

2.5.Data Analysis

2.5.1 Developing the model

To make the model for the potential distribution of Egyptian Vulture (Neophron percnopterus, we applied the maximum entropy algorithm from Maxent version 3.3.3k (downloaded on 23 June 2014 on ( to present data only, and this made a model of the habitat suitability of the species based on the climatic variables used as a probability distribution (Phillips et al., 2006).

2.5.2 Maxent

Maxent is a general-purpose method that is deliberate for creating predictions distribution modeling of species from incomplete information. Meanwhile, lately it has been commonly used as a general approach for presence-only modeling of species distribution. The approach of Maxent is discovering the probability distribution of maximum entropy to estimate a target probability distribution of species with restrictions that characterize lack information that is closest to uniform. The information of target species distribution is present as real value variables, which are called features, and each restriction are estimated value of each feature should equal its observed average. In the application of Maxent with presence-only species distribution modeling the pixels of the survey area are constructed the space and the pixels with species occurrence records constitute the sample point with environmental layers(Phillips et al., 2006). We used Maxent program version 3.3.3k, which is developed by S. Phillips and colleagues (for free download see: We run the Maxent model with default settings: auto features, maximum iteration 500, and converge threshold (10-5), while random test percentage was set to 25%, which is used 9 data point for modeling in this study. The program carried out the random test percentage automatically where 75% of the occurrence data points were randomly selected to train the model, which are 29 data points in this study, and the remaining 25% occurrence data points were used to test against the general model. The program selected suitable regularization values, included to reduce over fitting, automatically. Choosing environmental variables were also selected automatically. Maxent gives a probability of existence to each cell in the survey area. Since these probabilities must sum to 1, each cell’s probability is usually very small, making model output hard to interpret.

2.5.3 Maxent output data format

Maxent have three types of data output format, raw, cumulative and logistic formats(Phillips & Dudík, 2008). Raw values are initial output of Maxent and they must sum to 1, interpretation is difficult, because of they are very small for each data point. But another easier way for interpretation of Maxent output by using cumulative format, which is scale independent for location match to the probability of discovery of the species at that place which is match or lower likelihoods and their scores range from 0–1, it is not necessarily proportional to probability of presence (Phillips & Dudík, 2008; Phillips, Dudík, & Schapire, 2004). However, the more recent option for Maxent program is logistic format it provides evaluations of the probability of occurrence as predicted by involved environmental variables. It means that large changes in resulting output values will resemble improved with large differences in fitness. Also. The logistic output scopes score from 0–1. Then, the logistic format allows for easier and potentially more accurate interpretation over the other approaches. All output could be imported into a GIS map probability distributions(Phillips & Dudík, 2008).

2.5.4 Model evaluation

The receivers operating characteristic (ROC) curves were used to evaluated the model. It is a standard, threshold-independent method for model evaluation; these have been broadly used for this drive and are one of the Maxent’s output options (P Anderson et al., 2006; Phillips et al., 2004). Furthermore, Maxent generates a single measure of model produce, the area under the curve (AUC), which gives an indication of how the model is accurate and how the model is good in fit based on the predicted area.

Values close to 1 indicate models that only predict the points used to create the model, and these are referred to as over fitted; and values 0.5 indicate that the model is not better than expected at random(P Anderson et al., 2006). Additionally, Maxent achieves a jackknife test to evaluate the contribution of each environmental variable to the model, which is used to recognize the influence of each variable on the gain of the model, even if they are correlated, they are used in isolation or are omitted. All presence/absence models were used additional methods for evaluation the model, which is located in the analysis of the two types of prediction errors: false negatives, which are an omission error under prediction rate and false positives also are a commission error upper prediction rate (Ward, 2007). Maxent algorithm analyzes an omission rate test and training data. Omission rate indicates the fraction of test areas that drops into pixels, which is the fraction of all the pixels that are predicted as suitable for the species. (P Anderson et al., 2006). For a good model implementation, the model should be having a high proportion predicted localities properly with low omission error (false negative rate) are necessity but not enough (Phillips et al., 2006; Ward, 2007). The result of Species distribution modeling was used to predicted suitable habitat in Iraq for Egyptian Vulture in areas for which observational data do not exist. The threshold of equal training sensitivity and specificity was used to differentiate between suitable and unsuitable habitat for Egyptian Vulture. The AUC is not good to evaluate the model in some cases because it is affected by the prevalence of data and study area(Lobo, Jiménez‐Valverde, & Real, 2008). But in my cases it is good because the recorded points were distributed in away that make the AUC a reasonable method to be used to evaluate the model.

Table 1. Showed environmental variables that might be used in Maxent for potential distribution modeling of Egyptian Vulture within Iraq. The abbreviations presented them in tables and the data sources from which they were calculated. All variables in the list were continuous except land cover, which was categorical. The resolution of all the environmental variables was 1×1 km.

Abbildung in dieser Leseprobe nicht enthalten

Table 2. Land cover of Iraq (ESA, 2009)


Excerpt out of 53 pages


Predicting suitable habitat of Egyptian Vulture (Neophron percnopterus) in Iraq, using Maxent model
Applied Ecology and Conservation
Catalog Number
ISBN (eBook)
ISBN (Book)
File size
2091 KB
predicting, egyptian, vulture, neophron, iraq, maxent
Quote paper
Mariwan Rahim (Author), 2014, Predicting suitable habitat of Egyptian Vulture (Neophron percnopterus) in Iraq, using Maxent model, Munich, GRIN Verlag,


  • No comments yet.
Read the ebook
Title: Predicting suitable habitat of Egyptian Vulture (Neophron percnopterus) in Iraq, using Maxent model

Upload papers

Your term paper / thesis:

- Publication as eBook and book
- High royalties for the sales
- Completely free - with ISBN
- It only takes five minutes
- Every paper finds readers

Publish now - it's free