Prediction of Climate Change Effects on Siberian Crane (Grus leucogeranus) Habitat Suitability by Using Ensemble Modeling in Asia Wetlands

The Siberian Crane (Grus leucogeranus) is the third rarest and the most endangered crane species in the world. This study aimed at predicting the effect of climate change on Siberian Crane habitat suitability of breeding range and wintering range in Asia Wetlands by using ensemble modeling under future climate scenarios before the year 2050. In this regard, we used 4 modeling methods, Surface Range Envelop (SRE), Random Forest (RF), Classification Tree Analysis (CTA) and Flexible Discriminant Analysis (FDA) to determine the relationships between the species occurrence and bioclimatic variables under the ensemble framework by using Biomod2 and R software. The results showed that the AUC values were greater than 0.9 and functioning of all models was excellent. The Temperature Seasonality and Temperature Annual Range in the breeding range and Temperature Seasonality and Mean Temperature of Coldest Quarter in the wintering range had the most important role for habitat suitability of this species and respectively 105.64% and 136.27% changes was justified in Siberian Crane habitat suitability. Under RCP2.6 and RCP8.5 climate scenario for Siberian Crane breeding and wintering range, it is possible that climate change will cause a 100% loss of suitable habitat in West Sibera, and a 25.28% loss in Iran and China by 2050. The results of this study can be used in planning and conservation of crane species.


Introduction
Birds are highly sensitive to climate changes, with a strong ability to move; they are often considered as pioneering indicator species of climate change on animals (e.g., Stephens et al. 2016). Bird scientists had been focused on the impacts of rapidly changing climate conditions on the wild birds population over the past few decades. Global climate change has impacted the survival of many bird species (e.g., Crick 2004;Wu et al. 2012). Geographical distribution, breeding ecology and population dynamics of birds have changed to adapt to global climate change (e.g., Bård et al. 2005;Gasner et al. 2010;Chen et al. 2011). Climate change may be a long-term threat to breeding places, with changes in the freeze layer causing lakes to expand and the loss of islands, peninsulas and lowlying coastlines (e.g., Harris 2008;Van Impe 2013).
Species distribution models (SDMs) based on ecological niche limitation have been widely used to predict potential consequences of global changes in species distribution (e.g., Araújo et al. 2011;Maiorano et al. 2011;Early and Sax 2014). Recently, based on the ISI's Essential Science Indicators of species distribution modeling was rated as one of the best five research fields in ecological and environmental sciences (e.g., Renner and Warton 2013). Modeling and mapping of species distributions are essential for effective management and conservation of goal species (e.g., Lewis et al. 2017). Therefore, bioclimatic modeling based on ecological limitations imposed by climatic factors seems to be among the most practical and useful methods to make efficient predictions of distributions and extinction of threatened populations (e.g., Lovejoy and Hannah 2005).
The Grus leucogeranus is the third rarest and the most endangered crane in the world. The total population is approximated at 3,600-4,000 birds, almost all the population is in the East Asian Flyway. The Siberian Crane disperses across vast, inaccessible wetlands so double counting (or missing some cranes) may be impossible to avoid. Accordingly, the full count depends on the days when the birds are highly concentrated or when several counting parties can be coordinated with each other. This population estimate is based on years of synchronized winter counts at Poyang Lake (e.g., Li et al. 2012), supplemented by years of migration counts at Momoge where in recent years almost all Siberian Cranes have assembled on one big wetland. The Siberian Crane is propounded the most endangered crane species (e.g., Mirande and Harris 2019). Siberian Crane is a migratory bird species listed as Critically Endangered (CR) by the IUCN red list and also listed in CITES Appendix I (IUCN 2020). Assessment of the effects of climate change on hydrology and ecosystem performance for selected sites, in the context of impacts from current human activities require key research and monitoring of crane species (Mirande and Harris 2019). In this paper, we applied a prediction of climate change effects on Siberian Crane habitat suitability of breeding range and wintering range by using ensemble modeling in Asia under future climate scenarios before the year 2050.

Study Site
Asia is largest and most populous continent on the earth, located primarily in the Eastern and Northern Hemispheres. It stocks the continental landmass of Eurasia with the continent of Europe and the continental landmass of Afro-Eurasia with both Europe and Africa. Asia covers an area of 44,579,000 square kilometres (17,212,000 sq mi), about 30% of Earth's total land area and 8.7% of the Earth's total surface area. Asia is globally important for having high biodiversity, and along with South America has the highest rates of species richness (MacKinnon 2002). Of the world's 25 identified biodiversity hotspots, seven are in Asia, covering the entire Association of Southeast Asian Nations (ASEAN) area, plus the Western Ghats of India, Sri Lanka, southwest China and the eastern Himalayan countries of Nepal, Bhutan and India (Myers et al. 2000). The Hengduan Mountain region of China is the richest temperate ecosystem in the world (MacKinnon 2002). Southeast Asia contains the highest average proportion of country endemic bird and mammal species and second largest vascular plant species compared to other tropical areas (Sodhi et al. 2010). One out of 8, or a total of 324, of the 2,700 Asian bird species are globally threatened, including 41 that are critically endangered, 66 endangered and 217 vulnerable (Bird life International 2003). An additional 317 near threatened species are close to specifying as globally threatened, giving 664 species of conservation concern in the avifauna of Asia. Bird species experience a significant steep increase in extinction risk in Southeast Asia due to overhunting, habitat degradation, climate change and etc. (SCBC 2010).

Siberian Crane Records
The Siberian crane, also known as the Siberian white crane or the snow crane, is a bird of the Gruidae family, the cranes. They are prominent among the cranes, adults are nearly all snowy white, but for their black primary feathers that are obvious in flight. There are two populations in the Arctic tundra of western and eastern Russia, with the eastern breeding population migrating to winter in China and the western nonbreeding population migrating to winter in Iran and formerly India (Fig. 1).
We collected Siberian Cranes presence at 7 different sites for our field surveys. Our main sources were the Department of Environment (DoE) Fereydunkenar wetlands and the downloaded of Polygons(shp) data from the breeding range and wintering range sites from IUCN website and International Crane Foundation (e.g., Sadeghi Zadegan, personal comm. 2016; IUCN 2019; DOE 2021).

Climate Change Scenarios
Representative concentration pathways (RCPs) are the two independent pathways of greenhouse gas concentration trajectories developed by the Intergovernmental Panel on Climate Change (IPCC). Under the suppositions of the concentrationdriven RCPs, the average global surface air temperature will increase from 1986 to 2005 to 2045-2065 by 0.4-1.6 °C (RCP2.6) and 1.4-2.6 °C (RCP8.5) (IPCC 2013) respectively.

Modeling
In this study four modeling methods were used to evaluate the impacts of bioclimatic variables on the distribution of Siberian 1 3 Cranes: Surface Range Envelop (SRE), classification tree analysis (CTA), flexible discriminant analysis (FDA), and random forest (RF). Based on presence-only model used (e.g., Elith et al. 2006;Elith et al. 2010;Barbet-Massin et al. 2012), these four models need simulated pseudo-absence data on places where Siberian Crane are absent. To avoid creating pseudo-absence points that would be placed within or near the presence points, we used a random sampling plan which excluded the buffer zone of 1 and 9 km around the presence points. Using this buffer zone and following the methods of Barbet-Massin et al. (2012) and Senay et al. (2013), we randomly produced a set of pseudo-absence points using the Create Random Point tool in ArcGIS 10.3. We using the biomod2 package (Thuiller et al. 2016) in R v. 3.1.2 (R Development Core Team 2014). We applied the variable significance function to track changes in model performance by adding the reduction in model statistics with the addition of each variable into the model (Kuhn 2008). The output was that modelling probabilistic map was showing habitat suitability for Siberian Crane under current conditions of Bioclimatic variables. The ensemble map was made by the averaging of the projections made by different models (e.g., Marmion et al. 2009;Ashrafzadeh et al. 2018). We used variables of the ensemble map and its underlying ensemble model for projecting the potential distribution of Siberian Crane by the year 2050 under two RCPs of the Community Climate System Model version 4 (CCSM4). In addition, ensemble forecasting was performed by using a proportional weighted average  Temperature Seasonality (standard deviation *100) bio7 Temperature Annual Range (BIO5-BIO6) bio11 Mean Temperature of Coldest Quarter bio12 Annual Precipitation of each model's predictions based on the area under the curve (AUC) scores. We knew the critical levels of predictor variables to divide habitats into the two classes of suitable habitats and unsuitable habitats based on receiver operating characteristic (ROC) criteria (Sangoony et al. 2016) and used differences in each class to generate habitat suitability maps. From habitat suitability maps, we identified habitat gains and losses that may be caused via climate change by 2050 under two RCP scenarios in each Global Climate Model (GCM) (Marmion et al. 2009). We used ArcGIS 10.3 to map and study modeled Siberian Crane distribution and suitable habitats. To validate models, we randomly selected 80% of 5300 presence points as a training set and 20% of these 5300 points as a test set. In order to estimate the area of Siberian Crane map, we overlapped current habitat suitability maps with existing presence data in Asia and estimated their percent overlap in ArcGIS 10.3. We estimated model accuracy with two different criteria: (1) AUC as a single measure of model accuracy which is independent of thresholds and outbreak (e.g., Fielding and Bell 1997;Manel et al. 2001) and (2)

Current Range
Figure 2 Shows, that the Siberian Crane absences/presences in Asia Wetlands. According to the ensemble model, 6.04% of the study area is suitable breeding range for the Siberian Crane at present in West and East Siberia Wetlands. Also 1.56% of the study area is suitable wintering range for the Siberian Crane at present in China (Po Yang) and Iran (Mazandaran Fereydunkenar) Wetlands.

Breeding Range
The models were good to excellent according to their AUC > 0.900 and also were rated as good to excellent by TSS > 0.63 (Table 2). The accuracy of RF was the highest (AUC = 1), followed by CTA (0.991) and FDA (0.983).
Variables of Temperature Seasonality (Bio4) and Temperature Annual Range (Bio12) had the greatest relative importance on habitat suitability of this species and they were justified 105.64% positive changes in Siberian crane habitat suitability of breeding range (Table 3).
The ensemble models showed that until 2050, suitable habitats for the breeding range of the Siberian crane under RCP2.6 and RCP8.5 climate scenarios may be 5.92% and 5.71%, respectively, suitable habitats for the breeding range of the Siberian crane in East Siberian wetlands. And under the same scenarios, 94.42% and 91.90%, the new suitable habitat of the Siberian crane, 5.51% and 5.20%, respectively, and the habitat reduction in the western Siberian wetlands of its suitable climatic habitats due to climate change factors. (Table 4) ( Figs. 3, 4, 5 and 6).

Wintering Range
All models were good to excellent according to their AUC > 0.900 and also were ranked as good to excellent by TSS > 0.63 (Table 5). The accuracy of RF was the highest (AUC = 1), followed by FDA (0.944) and CTA (0.934).
Variables of Temperature Seasonality (Bio4) and Mean Temperature of Coldest Quarter (Bio7) had the greatest relative importance on habitat suitability of this species and they were justified 136.27% negative changes in Siberian Crane habitat suitability of wintering range (Table 6).

Discussion
Climate change will make a strong impact on the distribution of breeding and wintering Siberian Crane in Asia Wetlands before the year 2050 in accordance with Van Impe (2013) and Gasner et al. (2010). Climate change may lead to the loss of 2-25.38% of currently suitable areas in breeding range and wintering range but also may transform 0.72-5.51% of unsuitable habitats in breeding range and wintering range into suitable habitat. In some areas in the northeastern Siberia, the sizes of lakes will increase due to permafrost melting and wave action over the next half century, with open waters replacing breeding habitats for Siberian Cranes in accordance with  Hansbauer et al. (2014a, b) and Mirande and Harris (2019). Thus, in agreement with past studies, the cascading effects of climate change will place Siberian Crane at a greater risk than expected from habitat depletion alone. Climate change is destroying breeding habitat through erosion of the edges of the lake by waves and ice melt will increase the surface area  Mirande and Harris (2019). The fluctuating water conditions typical of semi-arid areas are expected to increase with climate change in Siberian Cranes habitat. Our findings are also supported by Mirande and Harris (2019). We found that climate change will be moved Siberian Crane range to parts of East Siberia in breeding range and parts of North Pakistan (Tarbela Dam), Iran (Gilan) and China (East of Po Yang Hu) in wintering range in accordance are with Mirande and Harris (2019).
The populations of Siberian Cranes are breeding in the northern Siberia and wintering in Iran (only 1 crane), India (crane are no longer) and China. Shooting, habitat degradation and climate change effected the destruction of the small Western Asian population (nine birds in 1996) that wintered on Fig. 6 The distribution of breeding Siberian Cranes using ensemble models under RCP8.5 by 2050 the Caspian lowlands of Iran (Fereydoonkenar) and the Central Asian population (four birds in 1996) of Siberian Cranes at Keoladeo National Park in India. In the Western Asia population the number of birds has decreased and only one Siberian Crane, named 'Omid' winters in Iran of Fereydoonkenar each winter from 2006 to 07 to 2020-22. During the winter of 2002-03 only one pair was observed at Keoladeo National Park in India, and they have not been reported since. Siberian Cranes did not return to Keoladeo in the following winter. Our findings are also supported by (e.g., Sadeghi Zadegan et al. 2009;Vuosalo Tavakoli 2014;Sadegh Sadeghi Zadegan, personal comm. 2016;Mirande and Harris 2019;DOE 2021). There are some Siberian Cranes on the West Siberia breeding range and at migration sites, offer the existence of unknown additional wintering grounds for the central and western flyways. Also, this is in agreement with other scientists (e.g., Shilina 2008;Shilina 2013;Rusanov et al. 2013;Rusanov 2014;Wetlands International 2014). On the wintering ranges of the Siberian Crane in Mazandaran Province (Fereydunkenar) of Iran, there is gun shooting of waterfowl, potentially putting cranes at risk (Sadeghi Zadegan 2011).
Climate change will create new suitable habitats of breeding range and wintering range in the northern and eastern parts of Russia and China where there are suitable climate conditions for Siberian Cranes. However, some populations in the west Siberia in breeding range and Iran and India in wintering range of the Asia Wetlands may become locally completely extinct. However, a significant portion of habitats, except for the East Siberia 1 3 in breeding range and China (Po Yang Hu) in wintering range of the Asia Wetlands, will perhap be unaffected by climate change. Even if climate change may make some suitable habitats, it is attended to degenerate the quality of Siberian Crane habitats in breeding range ,wintering range, summering range and critical stopover sites by strongly changing local practices and land use patterns. Dynamics of Siberian Crane populations in breeding range and wintering range are attached to climate change by direct effects on thermoregulatory and energetic processes and indirectly through climate effects on vegetation dynamics (phenology), water, food, population dynamics, and ecosystem ecology (White et al. 2018). Our results show that the Temperature Seasonality and Temperature Annual Range in the Breeding Range and Temperature Seasonality and Mean Temperature of Coldest Quarter in the wintering range the highest portion to had the greater effect on habitat suitability for Siberian Cranes. Effect of climate change on wintering range of Siberian Crane in China shows that mean temperature in wintering period will increase, and they are importance of considering the impacts of long-term climate variables on the habitat and population size of endangered birds, Siberian Crane. Extended drought conditions in the area can result in increased bird fatality, because it reduces water and food resources (Butler et al. 2014). Due to the threat of climate change Siberian cranes will miss breeding ranges, is most obvious with the critically endangered in the western Asia population of Siberian crane. Polar areas will see large changes in temperatures and weather patterns, that are affecting habitats critical for cranes. Also, this is in agreement with other scientists (e.g., Harris 2008;Beilfuss 2012). The key places where Siberian Cranes are most at risk include wintering ares at northeast China, especially Poyang and Dongting Lakes, Siberian Cranes at their migratory stations in southern China and their wintering fields in the mid-Yangtze River Basin and breeding areas at southern Yakutia; and Siberian Cranes at key regions in northeastern China, stopover sites at Yakutia, Lena basin in eastern Siberia and the Ob basin in Western Siberia, Siberian Cranes of Western and Central populations are key places currently most at risk in accordance with (Mirande and Harris 2019). Siberian Cranes wintering at poyang Lake in China face habitat loss due to long-term climate change as well as short-term threats. Poyang is the wintering range for almost all the world's Siberian Cranes (Barzen 2008;Hou et al. 2020Hou et al. , 2021Li et al. 2020). We found that climate change in the future will lead to important contracting of habitats suitable for breeding of Siberian Cranes and may cause a serious threat to the species' existence. The greatest effects of climate change on cranes are related to changes in water availably (Harris 2010). Our results highlight the importance of paying attention to the effects of long-term climate variables on the population size of the endangered birds, the Siberian Cranes. More attention needs to be paid to improving significant measures due to protecting habitat, managing water levels, etc. to protect cranes based on climate change (Xu et al. 2018).
The models used in our study predicted that climate change will have an effect on the distribution of Siberian Crane in Asia Wetlands of breeding range and wintering range before the year 2050.

Conclusion
Climate change may lead to the destruction of currently suitable regions in breeding range and wintering range and may turn unsuitable habitats in breeding range and wintering range into more suitable condition. Our study indicate that the effects of climate change will put Siberian Crane at a bigger risk than expected from the habitat downfall. The undulating water status typical of semi-arid areas are expected to increase with climate change within the range of the Siberian Cranes habitat. Climate change, unprotected hunting and habitat destruction could cause the extinction of the small Western Asian population wintering in Iran, India and China Wetlands, as well as population breeding in the Arctic tundra of western Russia Wetlands. Climate change will provide the development of new suitable habitats of breeding range and wintering range in the northern and Eastern areas of the Russia and China Wetlands with suitable climate conditions for Siberian Crane. However, it is not known whether cranes can adapt to these areas so quickly that they can survive in the wild.