Hotspots of (sub)alpine plants in the Irano‐Anatolian global biodiversity hotspot are insufficiently protected

The mountainous regions in SW Asia harbour a high number of endemic species, many of which are restricted to the high‐elevation zone. The (sub)alpine habitats of the region are under particular threat due to global change, but their biodiversity hotspots and conservation status have not been investigated so far.

Identifying threatened areas with both rich biodiversity and a high proportion of range-restricted species but poor conservation management (conservation gaps) is very important in conservation biology in general (Myers, 1988;Scott et al., 1993) and for the alpine zone in particular.
A suitable study region are the mountain ranges of SW Asia, where three global biodiversity hotspots (Mittermeier et al., 2011) and four biogeographical regions (Zohary, 1973) meet, resulting in a high biodiversity (Firouz, 2005;Manafzadeh et al., 2017;Noroozi, 2020;Zohary, 1973). By far the largest part of the alpine zone of SW Asia is located within the Irano-Anatolian global biodiversity hotspot, a region very rich in endemics but at the same time suffering high anthropogenic activities (Mittermeier et al., 2011;Noroozi, Naqinezhad, et al., 2019).
Several studies worked on biodiversity hotspots and their conservation status in this region (Ambarli et al., 2016;Eken et al., 2016;Fayvush et al., 2013;Memariani et al., 2016;Mohammadi et al., 2021;Noroozi et al., 2021;Noroozi, Naqinezhad, et al., 2019;Türe & Böcük, 2010;Yusefi et al., 2019). It has been shown that the plant diversity hotspots across all elevational zones of the Iranian Plateau are insufficiently protected and suffer from significant conservation gaps (Noroozi, Naqinezhad, et al., 2019). However, despite the very high endemic richness in the alpine flora and its strong influence on the bioregionalization of the region , hotspots of alpine species are mainly unknown and their conservation status are largely missing.
Most commonly used indices identify biodiversity hotspots by focusing on taxonomic diversity (Brooks et al., 2006;Linder, 2001;Reid, 1998) including richness in endemics or threatened species per spatial units (Kier et al., 2009;Norman, 2003). Additional insights when identifying biodiversity hotspots and conservation priorities can be gained by using phylogenetic diversity (Cadotte et al., 2010;Forest et al., 2007;Omland et al., 2008;Rosauer et al., 2009). Arguably, the most comprehensive approach towards identifying biodiversity hotspots is using several indices each emphasizing different aspects, such as phylogenetic distinctiveness or range size; such indices include species richness, endemic richness, range-restricted endemic richness, range-rarity richness, species phylogenetic diversity and endemic phylogenetic diversity (Cadotte et al., 2010;Crisp et al., 2001;Linder, 2001;Rosauer et al., 2009;Williams, 2000). Where several indices are congruently high, greater confidence can be placed in the resulting delineation of hotspots.
In this study, we want to identify (sub)alpine biodiversity hotspots within the Irano-Anatolian global biodiversity hotspot ('hotspots within a hotspot'; Cañadas et al., 2014) and their conservation gaps. To this end, we apply six biodiversity indices that consider either all species or only endemic ones. Specifically, we want to (i) test the hypothesis that the results of different biodiversity indices are highly congruent, as the proportion of endemic species in the alpine habitats of this region is very high; (ii) infer the locations of (sub)alpine plant diversity hotspots of the Irano-Anatolian global biodiversity hotspot and (iii) identify the conservation gaps, that is, hotspots that are not or insufficiently protected.

| MATERIAL AND ME THODS
The study area covers the high mountains of SW Asia (Turkey, Armenia, Iran, NE Iraq and S Turkmenistan) comprising nine major mountain ranges (Figure 1). This area is located at the intersection of four phytogeographical regions (Mediterranean, Irano-Turanian, Euro-Siberian and Saharo-Sindian; Zohary, 1973) and three global biodiversity hotspots (Mediterranean, Irano-Anatolian and Caucasian; Mittermeier et al., 2011). The climate of the region is very heterogeneous with a wide annual precipitation range (lower than 50 mm p. a. in the lowland deserts to more than 2000 mm p. a. in temperate humid forests); however, the (sub)alpine habitats are mainly characterized by continental climate with a Mediterranean precipitation regime, and in general, the precipitation increases from low to high elevations and decreases from west to east and from north to south (Djamali et al., 2011;Noroozi, 2020).
This study focusses on vascular plant species of the (sub)alpine belt defined as regions above 2300 m a.s.l. for Taurus Mountains, Pontic Mountains and the Armenian Mountains, above 2500 m a.s.l.
for Alborz, the Azerbaijan Plateau and Kopet Dagh-Khorassan, above 2700 m a.s.l. for Zagros, and above 3000 m a.s.l. for the Yazd-Kerman Massif. These elevation thresholds were set according to the cut-off for subalpine-alpine habitats recognized by previous publications (Fayvush & Aleksanyan, 2020;Noroozi & Körner, 2018;Parolly, 2020). Distribution data of all (sub)alpine species were compiled from various floras of the region (Assadi et al., 1989(Assadi et al., -2018Davis, 1965Davis, -1985Rechinger, 1963Rechinger, -2015 with a minimum spatial accuracy of 0.25° and were supplemented with extensive own observations (JN in Iran and EV in Armenia). The species were categorized as endemic when restricted to the study area, or subendemic when at least 80% of the species range was located within the study area.
We aimed at identifying hotspots of biodiversity at high elevations within the study area. As patterns in biodiversity are scaledependent (Anderson, 1994;Crisp et al., 2001;Major, 1988), we selected a cell size of 0.5° × 0.5° (roughly 59 × 44 km) to match the spatial resolution of the available occurrences. Six quantitative indices of species richness and endemism were calculated. Species richness (SR) and endemic richness (ER) are commonly used indices in biodiversity studies (Crisp et al., 2001;Linder, 2001;Williams, 2000) representing the total number of species and the number of (sub)endemic species in a grid cell, respectively.
However, hotspots of species richness do not necessarily coincide with hotspots of rare and threatened species (Hurdu et al., 2016;Orme et al., 2005). Therefore, we calculated two indices hardly affected by widespread species (Linder, 2001). Range-restricted endemic richness (RER) was defined as the number of endemic species present in not more than five grid cells (ca. 1% of the study area). Range-rarity richness (RRR), introduced by Crisp et al. (2001) as weighted endemism (for terminological aspects see , was calculated as sum of weighted species scores for a cell, where species scores are defined as 1 / number of grid cells occupied by the respective species (Williams, 2000).
Using the number of occupied cells and thus measuring area of occupancy  is preferred for alpine species, as their ranges often are discontinuous, so that measures of extent of occupancy will tend to overestimate ranges. Two indices of phylogenetic diversity were calculated using the ALLMB phylogeny provided by Smith and Brown (2018). This phylogenetic tree had to be pruned to the genus level as it does not include phylogenetic information at the species level for the majority of the study species. Furthermore, the tree was pruned to all genera in our data set and those of all endemics to calculate species phylogenetic diversity (SPD) and endemic phylogenetic diversity (EPD), respectively. The sum of the total phylogenetic branch length of the species within each cell was computed by applying the function 'pd' of the package 'picante'  of the statistical computing environment R (R Core Team, 2020). The correlation between all six indices was tested using a Pearson correlation test.
Cells with fewer than three species were excluded from all analyses as such cells are expected to contain the lowermost outposts of single (sub)alpine species rather than alpine habitats.
We defined the 5%, 10% and 20% of richest cells supported by any index as Hotspots (hotspots supported by at least one index), hereinafter referred to as 5% Hotspots, 10% Hotspots and 20% Hotspots, respectively. To this end, the six maps of single indices were merged into a single one and including all cells identified as Hotspots by at least one index. To identify Conservation Gaps (Hotspots not covered by protection), the map of Hotspots was overlaid with the distribution of three categories of nature reserves (National Parks, Wildlife Refuges and Protected Areas) across the study area. Thereby, only protected areas above 2300 m a.s.l. (i.e. the lowest elevational threshold used to define the (sub)alpine belt) were considered. To guarantee effective conservation, 10%-12% of each region should be under protection (IUCN, 2009); therefore, following Xu et al. (2017) any Hotspot cell, whose area is not covered by nature reserves by at least 10%, was defined as Conservation Gap. As the Marseille Manifesto emanating from the recent IUCN World Conservation Congress aims for 30% of the planet to be protected by 2030, we also report Conservation Gaps based on protection of maximally 30%, but focus our discussion on the Conservation Gaps identified with the more conservative threshold of 10% as this highlights the least protected areas of most urgent conservation. F I G U R E 1 Map of the study area (i.e. the Irano-Anatolian Plateau) comprising nine major mountain ranges (in yellow) of three global biodiversity hotspots (Mediterranean: brown; Irano-Anatolian: purple; Caucasian: green; after Mittermeier et al., 2011).

| RE SULTS
We compiled data of 1672 (sub)alpine species from 19,680 localities.
Within this set, 1267 species (76%) are (sub)endemics of the study region, 1014 species (61% of total species) are range-restricted endemics (i.e. found in a maximum of five cells) and 370 species (22%) are endemics found in a single cell (Appendices S1 and S2). Most of these 370 species are known only from their type localities.
Species richness ranged from 3 to 284 species per cell, ER ranged from 0 to 206 species per cell, RER ranged from 0 to 68 species per cell, RRR ranged from 0 to 37 values per cell, SPD ranged from 0 to 6155 per cell and EPD ranged from 0 to 4651 per cell (Appendix S3).
Pairwise correlations between indices ranged from R = .75 (RER with PD) to R = .98 (RE with RRR; Table 1) with a mean R of .88. Indices differing only in the used data set (all species vs. only endemic species) correlated very strongly (R = .96 for SR and ER, R = .95 for SPD and EPD). The values of all six indices were, thus, strongly correlated (Table 1), consequently revealing a congruent spatial pattern of Hotspots (Figure 2).
A total of 32, 53 and 98 cells were identified as Hotspots for the top 5%, 10% and 20% richest cells, respectively, using the threshold of 10% (Figure 3; Table 2). About 60% of these Hotspots were identified as Conservation Gaps (Table 2). Increasing the threshold to 30% (less than 30% of the surface area of a Hotspot is protected) resulted in 75%, 77% and 78% of the top 5%, 10% and 20% richest cells, respectively, being identified as Conservation Gaps. Moreover, only 22%, 18% and 16% of the (sub)alpine surface area of the identified Hotspots were covered by nature reserves for the top 5%, 10% and 20% richest cells, respectively ( Table 2).
Most of the Hotspots were found to be located in the Armenian Mountains, in Alborz and in the Hakkari Mountains (Figure 3), but their degree of protection differed. Whereas Hotspots in the Armenian Mountains are relatively well protected (60% of top 20% Hotspots, using the threshold of 10%), those in the Alborz and especially those in the Hakkari Mountains are poorly protected (42% and 14%, respectively, of the top 20% Hotspots, using the threshold of 10%; Figure 3). Additional Hotspots were located in the Pontic Mountains, in Zagros, in the Taurus Mountains as well as in Yazd-Kerman (Figure 3). There, 25%, 50%, 30% and 50%, respectively, of the top 20% Hotspots (using the threshold of 10%) are protected.

| DISCUSS ION
In general, a high proportion of alpine species (76%) is (sub)endemic to the study area, from those endemics nearly 80% are rangerestricted (i.e. present in maximally 5 cells), emphasizing the importance of these habitats for conservation biology (Brooks et al., 2006;Myers, 1988;Myers et al., 2000). Generally, patterns of biodiversity of alpine habitats have mainly been driven by habitat isolation, glacial history and environmental heterogeneity (Testolin et al., 2021).
The closest mountain range to the study area is the Caucasus, separated from the Armenian Highlands by a deep and wide lowland; other high mountains of the Irano-Turanian region in Central Asia, such as Hindu Kush and Pamir-Alai, are even further away. A second factor might be the isolation of alpine habitats within the study area. Compared to the alpine zones of the Alps and Caucasus, which are well connected and compact, alpine zones of the study area are strongly isolated and scattered (Figure 1) fostering allopatric speciation (Irl et al., 2017;Nosil, 2012;Steinbauer et al., 2016). As a third, not mutually exclusive factor, climate strongly determines phytogeographical regionalization (Djamali et al., 2012) and spatial patterns of plant diversity and endemism (Irl et al., 2015) in this area, suggesting that differences in climate contribute to this isolation. The Caucasus mountain range belongs to the Euro-Siberian phytogeographical region (Manafzadeh et al., 2017;Zohary, 1973), where the annual precipitation is high and evenly distributed throughout the year (Djamali et al., 2012;Egorov et al., 2020), but high elevations of the Irano-Anatolian region belong to the Irano-Turanian phytogeographical region with a continental climate and Mediterranean precipitation regime having wet-cold winters and warm-dry summers (Djamali et al., 2011(Djamali et al., , 2012. Moreover, the impact of glacial periods on the biodiversity of the study region was likely not as strong as in the high mountains of Europe, and most of the narrowly distributed species could have survived by elevational range shifts (Djamali, 2008;Noroozi et al., 2008). Finally, endemic richness in this region is positively correlated with environmental heterogeneity, which (using topographical heterogeneity as a surrogate) is high in the high elevations of the study area .
From the used biodiversity indices, RER (range-restricted endemic richness) and RRR (range-rarity richness) show the highest correlation. Both emphasize narrowly distributed species (Linder, 2001;Williams, 2000), which tend to be more threatened than widely distributed ones (Myers et al., 2000;Thuiller et al., 2005;Williams et al., 2008). Therefore, emphasizing Hotspots identified by RER and RRR is of particular relevance for conservation biology. On the other hand, as RER and RRR are highly congruent, we suggest that RRR is sufficient and that using RER, whose thresholds for range size of species are always arbitrary (Linder, 2001), is not necessary. Both SR (species richness) and ER (endemic richness) as well as SPD (species
phylogenetic diversity) and EPD (endemic phylogenetic diversity) are strongly correlated (.96 and .95 respectively), too, as expected given the high proportion of endemics and their role for bioregionalization of the study area . The cells richest for SR, ER, SPD and EPD are concentrated in three major mountain ranges: Central Alborz in northern Iran, Hakkari Mountains in southeastern Anatolia and the Armenian highland. Among those three, the Armenian highland stands out by having low RER and RRR values. This reflects that the region acts as a floristic corridor for nonendemic and subendemic species between the study area and the Caucasus mountain range.

| CON CLUS ION
This study shows that alpine Hotspots of SW Asia are insufficiently protected, although they are very rich in endemic and in narrowly distributed species. These alpine habitats are inside of global biodiversity hotspots (Figure 1), regions with very rich biodiversity and endemism and, at the same time, being under high pressure of anthropogenic activities (Mittermeier et al., 2011;Myers et al., 2000).
High elevations of the study region are particularly strongly impacted by overgrazing and global climate change, justifying emphasizing Conservation Gaps in (sub)alpine habitats, even though poor levels of protection of centres of plant endemism are not restricted to these habitats (Noroozi, Naqinezhad, et al., 2019). It should be considered that only four mountain peaks in the Irano-Anatolian Note: Total area -Total surface area of Hotspots above 2300 m a.s.l.; Protected area -Absolute and proportional surface area of the total area covered by natural reserves (national parks, wildlife refuges and other protected areas); Protected Hotspots -Number and proportion of Hotspots whose area is covered at least 10% by natural reserves; Conservation Gaps -Number and proportion of Hotspots, whose area is covered less than 10% by natural reserves.

ACK N OWLED G EM ENTS
The authors thank all botanists who collected floristic data from SW Asian high mountains and improved our knowledge on the plant biodiversity of the region. This study was financially supported by the Austrian Science Fund (FWF P31898 to JN and P28489 to GMS).

CO N FLI C T O F I NTE R E S T
We have no conflict of interest to declare.