Olea Ferruginea Classification Essay

On By In 1
Open Access This article is
  • freely available
  • re-usable


Predicting the Potential Distribution of Olea ferruginea in Pakistan incorporating Climate Change by Using Maxent Model

Uzma Ashraf 1,*, Hassan Ali 2, Muhammad Nawaz Chaudry 1, Irfan Ashraf 1,3, Adila Batool 4 and Zafeer Saqib 5


College of Earth and Environmental Sciences, University of the Punjab, Lahore 54590, Pakistan


Department of Zoology, University of the Punjab, Lahore 54590, Punjab Wildlife and Parks Department, Dera Ghazi Khan Region 32200, Pakistan


Strategic Policy Unit, Lahore Development Authority, Lahore 54770, Pakistan


Department of Space Science, University of the Punjab, Lahore 54590, Pakistan


Department of Environmental Science, International Islamic University, Islamabad 44000, Pakistan

Academic Editor: Marc A. Rosen

Received: 24 March 2016 / Accepted: 22 July 2016 / Published: 29 July 2016



The potential distribution of Olea ferruginea was predicted by Maxent model for present and the upcoming hypothetical (2050) climatic scenario. O. ferruginea is an economically beneficial plant species. For predicting the potential distribution of O. ferruginea in Pakistan, Worldclim variables for current and future climatic change scenarios, digital elevation model (DEM) slope, and aspects with the occurrence point were used. Pearson correlation was used to reject highly correlated variables. A total of 219 sighting points were used in the Maxent modeling. The area under curve (AUC) value was higher than 0.98. The approach used in this study is considered useful in predicting the potential distribution of O. ferruginea species, and can be an effective tool in the conservation and restoration planning for human welfare. The results show that there is a significant impact under future bioclimatic scenarios on the potential distribution of O. ferruginea in Pakistan. There is a significant decrease in the overall distribution of O. ferruginea due to loss of habitats under current distribution range, but this will be compensated by gain of habitat at higher altitudes in the future climate change scenario (habitat shift). It is recommended that the areas predicted suitable for the O. ferruginea may be used for plantation of this species while the deforested land should be restored for human welfare.


Correspondence: Tel.: +92-345-409-5182


bioclimatic; climate change; habitat shift; Jackknife test; Maxent model; Olea ferruginea

1. Introduction

O. ferruginea is a species of the family Oleaceae. Olea ferruginea is a synonym of Olea europaea subsp cuspidate. It is known as Indian olive and Kahoo in native language. From this species a number of different products can be obtained including fodder, firewood, edible fruits, and olive oil. Thus, olive is an important tree species from an economic point of view [1,2,3,4].

Past ecological studies highlighted the relation between the climatic indicators, occurrence, and abundance of different species [5,6,7,8,9,10,11,12]. The drastic change in species distributions or population status in an area is considered to be the result of land cover changes, human influence, and climate change [13,14].

Species distribution and habitat evaluation can be determined by different modelling techniques. [7,15,16,17,18,19,20,21,22,23]. The main purpose of this research is to predict the spatial shift of O. ferruginea in response to hypothetical projected climatic values for 2050. In ecological niche modelling (ENM) environmental parameters are compared within the geographical space. ENM techniques are based on statistical methods (e.g., generalized additive models (GAMs) and generalized linear models (GLMs)) and scientist-developed machine learning methods (e.g., artificial neural networks (ANNs) and maximum entropy (Maxent)) that are based on different statistical techniques [24,25,26,27,28,29,30].

If we relate ENM results of the habitat evolution of the species, there are fewer restrictions in terms of ecological variables, e.g., layers of environment and occurrence data where choice is available. Thus, unnecessary restriction should be avoided for unbiased results [7].

In a geographical space, ENM (ecological niche modelling) predictive high output values (high threshold) at grid cells shows relatively high probability of finding a species under suitable environment [14,31,32].

To evaluate the habitat and distribution of the Olea ferruginea in Pakistan, Maxent was used in this study. The study area includes the provinces of Pakistan (Punjab, Baluchistan, Khyber Pakhtunkhwa, and Sindh) and the area of Azad Jammu and Kasmir (AJK) except Kasmir territory under Indian control. The main objectives of the research are; (1) to understand the relation of O. ferruginea with reference to bioclimatic layers, digital elevation model (DEM), slope, and aspect; (2) to predict the current potential distribution of O. ferruginea in Pakistan by incorporating recent bioclimatic and topographic data; and (3) to project the distribution of O. ferruginea in Pakistan by incorporating Intergovernmental Panel on Climate Change (IPCC) climate change hypothetical projections (future bioclimatics).

2. Materials and Methods

Figure 1 shows the flow of methods which were adopted to predict the distribution of the O. ferruginea.

2.1. Data Collection

Species sighting data with longitude, latitude, and elevation were gathered by field surveys using quadrates of 10 × 10 m of Murree, Islamabad, Mansehra, Kohistan, Swat, Lower Dir, Bajur Agency, Malakand PA, Sibi, and D.I Khan (major districts where O. ferruginea can grow) as well as extracted from the literature [3,10]. Field survey was conducted while recording GPS points of O. ferruginea sightings. Before conducting the survey, literature was studied regarding Olea growing regions in Pakistan [4,10,33,34,35]. These geographical coordinates were helpful in locating the areas of O. ferruginea distribution. With the help of ArcGIS 10.2 (Esri, Redlands, CA, USA), the sighting point map was developed. A total of 219 sighting points were used for Maxent modelling, and are shown in Figure 2.

Three types of variables were used in this study, including topographical variables (DEM, slope, and aspect) and current and future bioclimatic variables. Literature review showed that a broad range of ecological inputs (environmental layers) can be used in predicting the potential distribution of Olea ferruginea by using habitat modelling, which in this case was Maxent [24,36]. DEM layer was downloaded from the SRTM web site [37]. Slope and aspect from the DEM data were derived by using “spatial analysis” tools of ArcGIS desktop and projected to WGS 1984 UTM zone 42 North [38,39]. Current and future (2050) bioclimatic variables for the study area were downloaded from the Worldclim website for the current conditions with a resolution of 1 km. Climate change can be predicted with fair degree of confidence in Pakistan up to the year of 2050 [40,41,42,43]. Bioclimatic layers were prepared by incorporating the past trends of temperature and precipitation. These are important to estimate the difference of current and the future patterns of climate and hence the species distribution. Future climatic projections were calculated using statistics by the Intergovernmental Panel on Climate Change (IPCC) 4th assessment data.

2.2. Data Processing

DEM, slope, and aspect were aggregated to 1 × 1 km pixel size to harmonize the dataset with bioclim datasets [39]. In Figure 1 the data processing method can be clearly seen. DEM and derived variables (slope and aspect) were used in the model because altitude is not going to change in the near future. After aggregating, all layers were converted to ASCII format by using model builder function of ArcGIS (extract by mask and raster to ASCII conversion tools were used). By using the SDM tool for ArcGIS the Pearson correlation was calculated. The Maxent model was also run using all variables to find out the most contributing variables for current predictions with an option of 10 replicates. The variables that have no impact on O. ferruginea prediction—based on current variables in the Maxent model—were ignored and the remaining variables were used for Maxent modelling [40]. Discriminant analysis was performed in Excel. Out of highly correlated (R > ±0.8) variables, one variable each was eliminated (Figure 3 and Table A1). Bio12 (annual precipitation), bio19 (precipitation of coldest quarter), bio17 (precipitation of driest quarter), bio11 (mean temperature of coldest quarter), bio18 (precipitation of warmest quarter), bio2 (mean diurnal range), bio14 (precipitation of driest month), DEM, bio7 (temperature annual range), bio4 (temperature seasonality), and aspect were used for further processing with the Maxent model, while other variables were ignored.

Maxent is a machine learning method. It works on the principle of maximum entropy defined as “how much choice is involved in the selection of an event” [40,42]. Maxent works efficiently with a low level of presence background data for animal or plant species, for prediction, as well as for projection of suitable habitat [24,44]. Maxent estimates the distribution of the species based on presence background data and imposes the restriction that the value of each expected variable should match its empirical average under this estimated algorithm [7,30]. For the current study, Maxent 3.3.3k version was used [45]. In settings, the random seed option was selected with a 10% of test sample size and 5000 iterations. Jacknife tests and the area under curve (AUC) with training and test data were also calculated in Maxent for evaluating the accuracy of the models. After running the model for predicting the current suitable habitat with selected input variables, prediction was projected for the future (2050) as well. For future predictions, ‘fade-by-clamping’ option was kept on so that over-prediction would be reduced [7,46]. For development of the species distribution map, ASCII outputs were converted to the raster and reclassified for area calculation.

3. Results and Discussions

The results are presented in three steps. First, the relative percentage contribution of predictor variables is given, followed by model performance evaluation on the bases of AUC values, and finally current and future hypothetical habitat suitability maps are presented and compared.

The Jackknife tests from Maxent show the relative contribution of each predictor variable for O. ferruginea prediction (Figure 3). The main variables for predicting potential distribution of O. ferruginea in Pakistan are annual precipitation (bio12) and precipitation of coldest quarter (bio19) for the current scenario with respective contributions of 38.6% and 20.5% (Table A1). Therefore, bio12, bio19, bio17 (precipitation of driest quarter), and bio11 (mean temperature of coldest quarter) have a greater contribution compared to other variables for the potential distribution of O. ferruginea. Bio18 (precipitation of warmest quarter) and bio11 (mean temperature of coldest quarter) are more important for permutation (Figure 3a). Figure 3b also shows that variables with high contribution factor have a higher AUC value. Aspect has low AUC value with less contribution, thus this is not important parameter for Olea distribution.

By applying the maximum entropy principle of Maxent, the maximum possible areas under receiver operating characteristic curves (AUC) test is 0.988 for current prediction and future projection [46,47]. The calculated AUC value shows that the results are significant (above 0.9) (Figure A1).

The model provides the species’ future distribution under hypothetical projected climatic values for 2050, compared to the climatic values under which the species currently occurs. The results calculated from present bioclimatics for the current prediction show that the total suitable area for O. ferruginea is approximately 41,449 km2, which is 4.83% of the total study area. In the future bioclimatic projection scenario (hypothetical), total distributional area of O. ferruginea is 19,192 km2, which is 2.23% of the total study area, as shown in Figure 4 (and Table 1). Species distribution maps were developed by using Maxent results. The map is categorized into four classes which are highly suitable (greater than 0.6 threshold value), moderately suitable (0.4 to 0.6), less suitable (0.2–0.4), and not suitable (<0.2) (Figure 4). For current and future classified maps, the threshold values are kept constant.

Figure 4c–f shows the zoom parts of the study area where major changes in the distribution were observed during current prediction and future projection.

Maxent results for current suitable habitats for O. ferruginea are predicted to be in northern Punjab (some parts of Khushab, Chakwal, Attock, and Rawalpindi) with low threshold value, in eastern to central and northwestern part of Khyber Pakhtunkhwa (KP) (Lower Dir, Swat, Kohistan, Karak, and D.I. Khan) with low to high threshold value, in northeastern parts of Baluchistan (Ziarat, Harnai, Mastung, and Kalat) and FATA (northwestern parts of Kurram Agency and western parts of Waziristan agency), AJK (Muzaffraabad, Hattian, Haveli, Bagh, Poonch, Kotli, and Sudhnoti) and in Islamabad (Figure 4c,e). The future projected species distribution map reveals that there is a significant decrease in the overall distribution of O. ferruginea, as can be seen from the calculated area (Table 1). Averaged future predictions from bioclimatic Maxent model of 2050 has shown a decrease in suitable habitat of Olea in Kurram Agency, central Khyber Pakhtunkhwa, northern Punjab, AJK, and Islamabad, and an increase in suitable habitat in northern Khyber Pakhtunkhwa (Figure 4d,f). So with future bioclimatics there will be a total decrease with a shift in Olea habitat toward higher latitude and altitude (Figure 4) [46,47,48].

Distributional maps are important sources for planning conservation strategy. Ground realities also indicate the area which is predicted by Maxent to be suitable for growth of O. ferruginea. The cutting of this species should be minimized because habitat fragmentation is the first step in degradation of any species.

4. Conclusions

This study presents occurrence data for the present and a model projection to a hypothetical future climatic scenario. Furthermore, future hypothetical projections of species response to environmental changes are difficult to ascertain. Therefore, there is a need to adopt proactive conservation planning measures using forecasts of species responses with respect to change in future environment.

Maxent software was used to predict the current and future potential distribution and habitat of Olea ferruginea in different climatic scenarios. The results reveal that under upcoming bioclimatic scenarios, there will be a moderate to high impact on the O. ferruginea distribution in Pakistan. There is a significant decrease in the overall distribution of O. ferruginea. It was also observed that in some areas O. ferruginea lost habitat in its present distribution range but shifted towards higher altitude and latitude in the future climate change scenario. It is recommended that the areas predicted suitable for the O. ferruginea with both current and hypothetical future scenarios be used for plantation of this species and the deforested land can be restored.

Author Contributions

Hassan Ali and Irfan Ashraf conceived and designed the experiments; Zafeer Saqib and Irfan Ashraf conducted the survey; Uzma Ashraf, Adila Batool, performed the experiments; Muhammad Nawaz Chaudry, Hassan Ali and Uzma Ashraf analyzed the data; Uzma Ashraf wrote the paper.

Conflicts of Interest

The authors declare no conflict of interest.


Olea[3] is a genus of about 40 species in the family Oleaceae, native to warm temperate and tropical regions of the Middle East, southern Europe, Africa, southern Asia, and Australasia.[2] They are evergreentrees and shrubs, with small, opposite, entire leaves. The fruit is a drupe. Leaves of Olea contain trichosclereids.[4]

For humans, the most important species is by far the olive (Olea europaea), native to the Mediterranean region, Africa, southwest Asia, and the Himalayas.[5][6]O. paniculata is a larger tree, attaining a height of 15–18 m in the forests of Queensland, and yielding a hard and tough timber. The yet harder wood of the black ironwoodO. laurifolia, an inhabitant of Natal, is important in South Africa.

Olea species are used as food plants by the larvae of some Lepidoptera species including double-striped pug.


Species accepted:[2][7][8]

  1. Olea ambrensisH.Perrier - Madagascar
  2. Olea borneensisBoerl. - Borneo, Philippines
  3. Olea brachiata(Lour.) Merr. - Guangdong, Hainan, Cambodia, Thailand, Vietnam, Peninsular Malaysia, Anambas Islands
  4. Olea capensisL. – Small Ironwood - Comoros, Madagascar; Africa from South Africa north to Ethiopia, Sudan, Zaire, Nigeria, Ivory Coast, etc
  5. Olea capitellataRidl. - Pahang
  6. Olea caudatilimbaL.C.Chia - Yunnan
  7. Olea chimanimaniKupicha - Chimanimani Mountains of Mozambique and Zimbabwe
  8. Olea cordatulaH.L.Li - Vietnam
  9. Olea dioicaRoxb. - India, Bangladesh, Myanmar
  10. Olea europaeaL. – Olive - Mediterranean, Africa, southwestern Asia, Himalayas; naturalized many other places
  11. Olea exasperataJacq. - South Africa
  12. Olea gagnepainiiKnobl. - Thailand, Laos
  13. Olea gambleiC.B.Clarke - Sikkim
  14. Olea hainanensisH.L.Li - Vietnam, Laos, Guangdong, Hainan
  15. Olea hochstetteriBaker - Kenya, Tanzania, Uganda
  16. Olea javanica(Blume) Knobl. - Philippines, western Indonesia
  17. Olea lanceaLam. - Madagascar, Mauritius, Réunion, Rodrigues Island
  18. Olea laxifloraH.L.Li - Yunnan
  19. Olea moluccensisKiew - Maluku
  20. Olea neriifoliaH.L.Li - Hainan
  21. Olea palawanensisKiew - Palawan
  22. Olea paniculataR.Br. - Yunnan, India, Indochina, Indonesia, Kashmir, Malaysia, Nepal, New Guinea, Pakistan, Sri Lanka, Australia, New Caledonia, Vanuatu
  23. Olea parvilimba(Merr. & Chun) B.M.Miao - Hainan, Vietnam
  24. Olea polygamaWight - India, Sri Lanka
  25. Olea puberulaRidl. - Peninsular Malaysia
  26. Olea roseaCraib - Yunnan, Thailand
  27. Olea rubrovenia(Elmer) Kiew - Borneo, Philippines
  28. Olea salicifoliaWall. ex G.Don - Assam, southern China, Indochina
  29. Olea schliebeniiKnobl. - Tanzania
  30. Olea tetragonocladaL.C.Chia - Guangxi
  31. Olea tsoongii(Merr.) P.S.Green - Guangdong, Guangxi, Guizhou, Hainan, Sichuan, Yunnan
  32. Olea welwitschii(Knobl.) Gilg & G.Schellenb. - central and eastern Africa from Ethiopia to Zimbabwe
  33. Olea wightianaWall. ex G.Don - India
  34. Olea woodianaKnobl. - South Africa, Swaziland, Kenya, Tanzania
  35. Olea yuennanensisHand.-Mazz. - China

Formerly placed here[edit]

List source :[7]


Wikimedia Commons has media related to Olea.
  1. ^ ab GRIN (April 4, 2006). "Olea information from NPGS/GRIN". Taxonomy for Plants. National Germplasm Resources Laboratory, Beltsville, Maryland: USDA, ARS, National Genetic Resources Program. Archived from the original on June 29, 2011. Retrieved May 20, 2011. 
  2. ^ abcKew World Checklist of Selected Plant Families, genus Olea
  3. ^Sunset Western Garden Book, 1995:606–607
  4. ^Flora of China v 15 p 295, 木犀榄属 mu xi lan shu, Olea Linnaeus, Sp. Pl. 1: 7. 1753.
  5. ^Kew World Checklist of Selected Plant Families, Olea europaea L.
  6. ^Altevista Flora Italiana, Oleastro, Olea europaea L.
  7. ^ abGRIN. "Species in GRIN for genus Olea". Taxonomy for Plants. National Germplasm Resources Laboratory, Beltsville, Maryland: USDA, ARS, National Genetic Resources Program. Archived from the original on September 24, 2015. Retrieved May 20, 2011. 
  8. ^"Name - Olea L. subordinate taxa". Tropicos. Saint Louis, Missouri: Missouri Botanical Garden. Retrieved May 20, 2011. 


Leave a Reply

Your email address will not be published. Required fields are marked *