Water availability limits brown bear distribution at the southern edge of its global range

Mehdi Ansari H., Arash Ghoddousi

Throughout the world, especially in densely populated landscapes, human activities have often driven wildlife from the most suitable habitats (Ehrlich and Pringle

2008). Characteristics such as wide-ranging behavior, need for expansive habitats, high metabolic demands, and natural low densities factor into making large car- nivores the most threatened species in the world (Ripple et al. 2014). These factors restrict large carnivores to in- accessible or remote areas in most of the world (Swenson et al. 2000). However, because carnivores are among the species most admired by the public, they have become iconic flagship species for conservation (Ripple et al.

2014).

The brown bear (Ursus arctos) is the most widely dis- tributed Ursid in the world and is a ‘Least Concern’ species in the International Union for Conservation of Nature Red List of Threatened Species (McLellan et al.

2017) on account of presence of healthy populations in North America, Eastern Europe, and parts of Asia. However, the status of bears in Iran (one of the south- ernmost extents of their range) is of greater concern (Gutleb and Ziaie 1999). The brown bear is the largest car- nivore of Iran, occupying forest and mountainous ecosys- tems of 2 main mountain ranges: Alborz and Zagros (Gutleb and Ziaie 1999). As a ‘flagship’ species (one chosen to represent broader conservation needs or goals; Noss and Cooperrider 1994, Noss et al. 1996, Miller et al. 1998), the brown bear has played an important role in conservation plans; however, it has not been widely studied in Iran (Gutleb and Ziaie 1999, Qashqaei et al.

2014, Ashrafzadeh et al. 2016). The species faces con- flict with humans over shared resources and shrinkage of habitat, which requires immediate conservation attention (Qashqaei et al. 2014, McLellan et al. 2017).

Predictive geographical models are useful tools for assessing the spatial distribution of species, planning wildlife conservation, assessing land use, and studying invasive species (Guisan and Zimmermann 2000, Manly et al. 2002, Thuiller et al. 2005, Pearce and Boyce, 2006). They are also widely used to estimate potential habitat quality (Nielsen et al. 2010), evaluate human–wildlife conflict (Le Lay et al. 2001), and plan species reintro- ductions (Adhikari et al. 2012). An understanding of the connection between spatial distribution of animals and their habitats is fundamental to conservation of threatened species and to conflict management (Lecis and Norris

2003). Food resources are vital factors affecting individ- ual growth of animals and abundance of populations (Car- bone and Gittleman 2002, Brasher et al. 2007). The abun- dance and availability of food resources are especially important to brown bears during the warmer months of the year, when their nutritional demand peaks as individ- uals accumulate fat to survive winter hibernation (Rode et al. 2006, Berland et al. 2008).

In their globally southernmost distribution extent in Fars Province, brown bears appear to be genetically dis- tinct from the other populations in Iran, forming a dis- tinct sub-clade (Ashrafzadeh et al. 2016). Brown bear habitat in this region is semi-arid, and both humans and bears compete for limited water and food resources. This demand for resource use brings brown bears into occa- sional conflict with farmers and poses a great risk of bear mortality due to lethal control measures (Qashqaei et al.

2014).

This study uses field surveys, remote sensing, Geo- graphic Information Systems (GIS; Imam et al. 2009), and maximum entropy (Maxent) approaches (Phillips et al. 2004) to achieve information about the habitat used by brown bear. Maxent is a general-purpose machine- learning method with a simple but precise mathemat- ical formulation that makes predictions from species- presence-only data and environmental variables (Phillips et al. 2006). It can be run with small sample sizes and is well-known for ease of model explanation, making it a suitable candidate for studying a rare species with minimum resources available in conservation (Phillips et al. 2006, Pearson et al. 2007, Elith et al. 2011). In this study, we model the probability of brown bear occurrence in the Fars Province during the feeding season, using a combination of field surveys and Maxent modeling, to better understand the environmental and anthropogenic variables influencing the species’ level of reliance on re- sources shared with humans.

Study area

The study area is located in northern Fars Province in the Pasargad Region of Iran and covers approximately

1,660 km2  (29°59 39   to 30°31 45 N and 52°17 35   to

52°18 58 E; Fig. 1). The area has 3 main mountain ridges oriented northwest–southeast. Elevation ranges from 1,600 to 3,200 m above sea level. Forests cover

73.6% of the study area, while 18.4% is rangeland, 6.4% is under irrigated agriculture, 1.3% is under unirrigated agriculture, and 0.3% is rocky area. Water resources, in- cluding springs, rivers, and livestock watering troughs, are scarce and play an important role in wildlife dis- tribution in the area. There are 2 rivers in the northern boundary of the study area: Sooten and Ojan. The study area is surrounded by paved and gravel roads and one of the national roads (average traffic: 5 vehicles/min) passes through it (Fig. 1). There are 53 villages at the margins and 4 villages and several nomadic tribes in- side the study area. The study area experiences cold to hot semi-arid climate with a maximum temperature of

33°C in summer and a minimum of 0°C in winter. Mean annual rainfall is 250 mm. Forest tree species include Acer monspessulanum, Amygdalus scoparia, Ficus sp., Pistacia atlantica, Pistacia khinjuk, Prosopis cineraria, and Ziziphus spina-christi; of these, P. atlantica and A. scoparia are the dominant species. Agricultural lands and orchards—primarily growing almond (Terminalia cat- appa), apricot (Prunus armeniaca), grape (Vitis vinifera), peach (Prunus persica), plum (Prunus domestica), and walnut (Juglans regia)—surround the study area. The other large wildlife species in the area are leopard (Pan- thera pardus), striped hyena (Hyaena hyaena), wild boar (Sus scrofa), wild (bezoar) goat (Capra aegagrus), and wolf (Canis lupus). There is no official protection status for the area and the land is mostly national, but approx- imately 8% of the study area land is privately owned. Moreover, no regulated hunting of brown bear occurs in Iran and poaching has seldom been reported within the area. However, occasional retaliatory killing of bears in response to crop damage, as well as mortalities due to road accidents, have been reported from the area.

Material and methods

Sampling of the entire study area was not possible given logistical and financial restrictions. To avoid bias in sampling only in areas with better accessibility or