Gazella arabica dareshurii: a remarkable relict population on Farur Island, Iran

The islands in the Persian Gulf are home to several species of gazelles, i.e., Gazella bennettii, G. subgutturosa, and a new subspecies of Mountain gazelles which was discovered on Farur Island and described for the first time in 1993 as Gazella gazella dareshurii. Later, phylogenetic analyses showed that the Mountain gazelles consist of two species: G. gazella and G. arabica. As the Farur gazelles are more closely related to the Arabian forms of the Mountain gazelles, this subspecies is regarded to be G. arabica dareshurii. Until now, the origin of this subspecies has been an enigma. Here, we used mitochondrial cyt b, two nuclear introns (CHD2 and ZNF618), and morphological data to address this question by investigating the taxonomic position of the Farur gazelles. The results show that this population is monophyletic and split from other G. arabica populations probably 10,000 BP. It is a natural relict population that was trapped on the island due to the rising sea levels of the Persian Gulf after the Last Glacial Maximum. Intermittent drought and flooding are suggested to be the main factors balancing population growth in the absence of natural predators on this monsoon-influenced island. Conservation actions should focus on preserving the natural situation of the island (cease introducing mesquite tree and other invasive species, stop building new construction and roads, and caution in providing water sources and forage), and possibly introducing individuals to other islands (not inhabited by gazelles) or to fenced areas on the Iranian mainland (strictly isolated from other gazelle populations) when the population reaches the carrying capacity of the island.

type material for the subspecies dareshurii [6]. Karami and Groves [6] took 23 measurements on each skull of which seven to nine were selected for multivariate analyses. They compared these skulls with other gazelle species, especially with other described subspecies of G. arabica, including G. a. cora (the common Arabian gazelle, later synonymized with G. arabica [14,15]), "G. a. erlangeri" (from the southwestern Arabian Peninsula, probably a pet gazelle and therefore placed in "" in the remaining text, see [16]), and G. a. muscatensis (from the Batinah coast of Oman). The results showed that although the Farur gazelles are similar to G. a. muscatensis in horn characters (lyrate horns in both sexes, wellformed horns with clear rings in females), they can be differentiated from them and other G. arabica taxa, with clearer separation of females than males in discriminant analysis. However, male skulls from Farur were not distinguishable from "G. a. erlangeri" in this analysis, despite the fact that they display very different horn shapes (straight horns in "G. a. erlangeri").
Based on these results, Karami and Groves [6] described G. g. dareshurii as "a subspecies of Gazella gazella (now G. arabica, see Bärmann et al. [14]) similar in size and horn shape to G.[a.] muscatensis but with longer horns in males; in both sexes the horns are broader across the base, the skull is much narrower, and the nasal bones are posteriorly narrower. Compared to G. [a.] cora, it differs additionally in its much smaller size, with shorter horns in the male but longer horns, broader at the base, in the female. Compared to G. [a.] erlangeri it differs primarily in its horns, which are outbowed, with the tips turned in, in both sexes. Special comparison with the much larger, straight-horned G. gazella from Israel and with the very small, also straight-horned, G. [a.] farasani from the Farasan Island, is unnecessary. " The origin of the Farur gazelles remains enigmatic until today and several hypotheses were put forward on how their presence on the island could be explained: (1) Karami and Groves [6], based on the pers. comm. with M. T. Moinian, suggested that eight individuals of unknown sex had been introduced to Farur Island in 1967 from Kavir National Park (NP) in central Iran. However, no signs of the existence of a possible source population could be found in Kavir NP, which is only inhabited by another gazelle species (G. bennettii) [6]. (2) Hemami and Groves [11] suggested that the gazelles from Farur Island might originate from some unspecified place on the Arabian Peninsula. Karami et al. [10] specified this suggestion and mentioned that the gazelles of Farur might be related to G. a. muscatensis, an enigmatic subspecies of G. arabica inhabiting the coastal plains of north-west Oman. However, both differ in fur coloration with the Farur gazelles having a pale sandybrown pelage rather than a deep chocolate-brown known from G. a. muscatensis. (3) Unconfirmed information from the local people says that Sheikh Oboud Moghuyehie introduced one male and one female to the island in the 1950s from Bandar Moghuyeh (habitat of G. bennettii) on the Iranian shores close to Farur (Fig. 1), but no documentation is known corroborating this hypothesis [17]. Thus, the question of the origin of the Farur gazelles remains unclear.
Here, we investigate the phylogenetic relationships and morphological similarity of the Farur gazelles with Arabian mainland G. arabica by using molecular and morphometric methods. We hypothesize that (1) G. a. dareshurii is a valid subspecies of G. arabica, and (2) Farur is the historic habitat that once covered the northern part of the Persian Gulf and this population of mountain gazelles was trapped on the island due to the rising sea level after the LGM.

Genetic analyses
Seventeen samples from Farur were successfully sequenced for one or more of the following markers: chromodomain-helicase-DNA-binding protein 2 (CHD2) (669 bp, ten samples), zinc finger protein 618 (ZNF618) (689 bp, nine samples), and cytochrome b (cyt b) (1140 bp, ten samples). For all three markers, only one haplotype was detected in all samples. Differences in the intron sequences of Farur gazelle compared with G. arabica and G. gazelle are shown in Table 1.
The phylogenetic tree of mountain gazelles shows G. arabica and G. gazella as sister species as expected from previous studies (posterior probability (PP) = 1). Within G. arabica, the samples from Farur Island form a monophyletic group (PP = 1) that is placed as sister to all other sequences from all over the mainland of the Arabian Peninsula ( Fig. 2 and Fig. 3). According to the molecular clock, the split between the two groups occurred around 0.7 Ma (Fig. 2).

Morphometric analyses Principal component analyses
In the first principle component analysis (PCA, Fig. 4a), including all specimens, two main components were found that together account for 74% of the variance of the data set. This analysis clearly separates males and females, so that separate PCAs were also conducted to explore the spread of the data for each sex. However, as only very few female specimens were available, both sexes were analyzed together in the discriminant function analysis (DFA).
In the PCA including only male specimens (Fig. 4c), four principal components were found (explaining 75% of the variance). The first and second components, together accounting for 60% of the variance, distinguish three groups of gazelles with only minimum overlap. The first component, mainly influenced by skull width and length, and occipital height, separates the larger G. gazella and G. a. arabica from the other groups. The second component, mainly influenced by horn width, horn tip distance, horn length, and braincase height, separates "G. a. farasani" from the other groups. Most of the G. a. dareshurii males are found in a cluster with "G. a. erlangeri" and G. a. muscatensis, but two specimens are closer to G. a. arabica in morphospace.  For the third PCA, including only females (Fig. 4d), the number of specimens was relatively low. We therefore excluded four measurements that showed very low extraction values (below 0.7) or were only influential in the fourth component in a preliminary analysis, to have a more appropriate parameter-to-specimens ratio. Three main components were found, together describing 86% of the variance of the data. The first component is influenced by measurements describing skull length and width, similarly as in males. But in addition, horn length and diameter, as well as braincase length and height, have a large impact. The second component is mainly influenced by horn distance at the base, maximum horn width, and facial length parameters (DFO, DFH). Together these two components separate the small and short-horned "G. a. farasani" as well as G. a. muscatensis from the other groups. G. a. dareshurii is situated in an intermediate position between G. a. arabica, "G. a. erlangeri", and G. a. muscatensis.

Discriminant function analysis
The DFA including all specimens (Fig. 4b) based on raw measurements had a success rate of 98.6%; only one G. a. arabica was misclassified as G. a. dareshurii. In cross-validation the success rate was 85.5%. The misclassifications involved almost all G. arabica subgroups, including three G. a. arabicas, one "G. a. erlangeri", and one "G. a. farasani" classified as G. a. dareshurii, as  [42]. Numbers above branches are posterior probabilities well as two G. a. dareshurii that were classified as G. a. arabica and "G. a. erlangeri", respectively (Table 2).

Molecular phylogenetic analysis
The phylogenetic analysis places G. a. dareshurii with G. arabica, as expected from previous analyses based on morphology. However, the Farur specimens are not nested within G. arabica, as would be expected if these gazelles were translocated from some Arabian mainland population in historic times, but form the sister-group to all other G. arabica (Figs. 2, 3). This implies that this taxon has split from the other Arabian gazelles a long time ago.

Morphometric analysis
The principal component analyses show that only "G. a. farasani" is clearly separate from the other G. arabica subgroups. These very small gazelles with very short horns inhabiting the Farasan Islands in the Red Sea were shown to be an ecotype rather than a subspecies of G. arabica [18], as high genetic admixture exists between the Farasan and the mainland gazelles (therefore the name is put in "" in this text). The only misclassification that occurred between "G. a. farasani" and G. a. dareshurii in the cross-validation of the discriminant function analyses involved a female (one female "G. a. farasani" classified as G. a. dareshurii), which might be caused by the very small numbers of females in our analysis (only three females each for "G. a. farasani" and G. a. muscatensis, only two females each for "G. a. erlangeri" and G. gazella). There seems to be no convergent "island morphotype" of Farur and Farasan gazelles that distinguishes them from the mainland gazelles.
A lot more similarity, and misclassification in the DFA, can be observed between G. a. dareshurii, "G. a. erlangeri", and G. a. arabica. As described by Karami & Groves [6], the Farur gazelles show the closest similarity to "G. a. erlangeri" in the PCA. The taxonomic status of  Table 5 "G. a. erlangeri" has recently been reviewed [16]. Based on genetic and morphometric analysis, the authors conclude that "G. a. erlangeri", only known from captive populations, is most likely a pet gazelle derived from a darker-coloured variant of G. arabica, possibly of the now extinct G. a. muscatensis. It is astonishing that the Farur gazelles, despite being isolated on the island for thousands of years, and being restricted to a very small population size (< 1000), have not evolved a distinct morphology. Instead, they mediate between the morphotypes of G. muscatensis and "G. erlangeri" on the one hand, and G. arabica on the other. They demonstrate that the morphospace of G. arabica does not display any major gaps between regional subgroups as were found in previous analyses based on available museum specimens.

The origin of G. a. dareshurii
During the LGM, the Persian Gulf was a river valley with a few hilly outcrops along the north-eastern rim [19]. Mountain gazelles might have inhabited these hills, as they are usually found in mountainous regions throughout the Arabian Peninsula [13,20,21]. These outlier populations would have been genetically isolated from the Arabian populations for a long time, as the fertile and well hydrated land between them acted as a barrier for dispersal.
After the LGM, about 14,000 BP, the rising sea levels lead to a flooding of the Persian Gulf [2]. During this process, Farur became an island about 10,000 BP [2,19]. We postulate that the Arabian gazelles on Farur survived on the island since that time, despite being restricted to an extremely small population size. In historical times, only 300-500 gazelles inhabited the island at any one time.
Possibly other outlier populations could have existed on other islands or in the mountain areas on the Iranian coast, but our extensive efforts to collect samples from all possible habitats did not reveal any other populations of G. arabica in Iran. So we can only speculate on why they did not survive. On the mainland and the larger islands, competition with other gazelle species, especially G. bennettii which is also adapted to desert conditions, might have led to the local extinction of G. arabica. On the smaller islands, the catastrophic effects of drought and flooding could have been too severe for the long-term survival of a small and isolated population of gazelles. On Farur, these natural threats did not lead to extinction, but now anthropogenic threats might negatively affect the Farur gazelles.

Threats for the Farur gazelles Natural threats
Although people were living on Farur in the past (< 1950), nowadays neither humans nor carnivores inhabit the island. Thus, the gazelles living on the island have no natural predators. Diseases and droughts were suggested to be the major threats to their survival [11]. Farur is under the influence of Indian summer (the middle of June-the middle of September) and winter (December-March) monsoon. Without doubt, the main natural threat is intermittent drought which was the main reason for the population decline in 1986 (low precipitation during the winter monsoon). In several years, also flooding events caused by heavy rain during monsoon were responsible for population declines where carcasses of gazelles were found in the valleys (Saman Ghasemi and Meisam Ghasemi, personal observations). However, diseases were so far not recorded to occur in this gazelle population,  probably due to the geographic isolation from other wild or captive bovids. But diseases could potentially have a large impact, as they most likely would infect the entire population. Therefore, intermittent drought (leading to food limitation) and flooding are possibly two main natural threats that might act as balancing factors in the absence of natural predators. They naturally restrict population growths and therewith prevent severe overgrazing of the island which otherwise could lead to a collapse of the gazelle population.

Inbreeding
It seems surprising that such a small population of gazelles that persisted on a tiny island without genetic exchange for thousands of years has not suffered from, or even gone extinct by inbreeding. However, it is not the amount of genetic diversity, but the absence of large amounts of strongly deleterious mutations that is responsible for the viability of small populations [22,23]. A genomic study on wild foxes inhabiting the Channel Islands in California for more than 9000 years has revealed that these populations show an extremely low amount of moderately to strongly deleterious mutations compared to mainland foxes [24]. This can be explained by genomic purging, i.e., the wiping out of moderately and strongly deleterious recessive mutations due to increased selection pressure, as these mutations are more often found in homozygosity in small populations [25]. We think that the situation on Farur is very similar, so that the absence of inbreeding depression can be explained by the very long isolation and the consistently small population, even before Farur became an island. The accumulation of mildly deleterious mutations that likely occurs in such small populations, as was demonstrated for Alpine ibex [26], still leads to a reduction of the fitness, but in an environment without predators this seems to have a low impact. In any case, genomic studies of these gazelles would be highly desirable to gain more information on genomic purging in island populations.

Anthropogenic threats
Farur gazelles adapted to the natural situation of the island, so every anthropogenic intervention can be a threat to the survival of the gazelles. Umbrella thorn (Acacia tortilis) is the main food source during the dry season, but a newly introduced tree, mesquite (Prosopis juliflora), now invasively occupies some part of the island [27]. This alien species is a potential threat for the Acacia and therefore also for the gazelles.
Recently, the Iranian navy has set up a camp on the island, increasing gazelle-human contact: in the dry season, gazelles tend to come close to the settlements in search of food and water. Several water reservoirs were built on the island to reduce the effects of drought for gazelles. Although the presence of the naval forces led to a complete stop of illegal hunting activities, the construction activities commissioned by the navy affect the natural habitat of gazelles. Iranian Department of Environment (DoE) authorized hunting for the first and last time in 2010 when wrongly considering the Farur gazelles to belong to G. bennettii, a common gazelle of the Iranian mainland.

Conclusion and implications for conservation
Farur gazelles are not only a subspecies of G. arabica, but also are a remarkable relict population estimated to have split from other G. arabica populations 0.7 Ma and survived on Farur in isolation (10,000 BP). Being trapped on the small island led to adaptation to the island's nature with drought and flood acting as balancing forces to regulate population growth in the absence of natural predators. Conservation actions are necessary for this relict population as its long-lasting separate evolutionary history might have led to the acquisition of genomic changes in adaptation to the specific island's requirements. It could therefore serve as an example of local adaptation [28], be used as a model for the assessment of evolutionary change and genomic purging [24,29], and represent a case study for biogeographical studies [30]. Farur gazelles are the only recorded population of G. a. dareshurii, so they should be treated as a separate conservation and management unit [31]. Therefore, conservation actions should aim at ensuring the survival of the population within its natural environment, and possibly introducing the species to other islands (not inhabited by gazelles) or the Iranian mainland (strictly isolated from other gazelle populations) when the Farur population increases to more than the island's carrying capacity.

Farur Island
Farur Island is located in the northern middle part of the Persian Gulf (Fig. 1). The shortest distance to the mainland is around 22 km, with Boustaneh as the nearest point on the Iranian mainland. The greatest length and width are 7.5 and 4.5 km respectively, and the area is 28.48 square kilometers with an elevation ranging from 0 to 140 m above sea level. The island is located between Kish Island in the west, Qeshm, Greater Tunb, and Lesser Tunb Islands in the east, and Siri and Abu Musa Islands in the south and southeast (Fig. 1). Farur Island is a protected area under the DoE since 1979. It is the only island in the Persian Gulf inhabited by Arabian mountain gazelle.
The island surface is uneven and hilly with several peaks on its central and western parts. The highest peak is 145 m and is located in the west of the island. Most of the island area is at an altitude of 50 m. Grass, shrub, and mostly trees that have adapted to the hot weather cover the island. Umbrella thorn (A. tortilis) is distributed on the whole island with the highest density in the valleys. Mesquite (P. juliflora) is an invasive species that recently arrived on the island [27].
The climate is tropical with seasonal mean temperatures of 27 °C in spring (March-May), 34 °C in summer (June-August), 29 °C in autumn (September-November), and 20 °C in winter (December-February) based on the Bandar Lengeh weather station  as the nearest synoptic weather station to the island. Annual precipitation is 133 mm and relative humidity is high during the year with over 90% on some days. The Persian Gulf is under the influence of Indian summer monsoon in the boreal summer from the middle of June until the middle of September, and Indian winter monsoon in the boreal winter between December and March with weaker, dry, and cold northeasterly winds compared to the strong southwestern monsoonal winds [32,33]. Total monthly precipitation  shows that December (28.20), January (34.05), February (26.52), and March (26.77) are raining months, and after April (6.17), the total monthly precipitation is less than one millimeter in May, June, and July. It seems that August (2.02) and September (1.02) are monsooninfluenced months in the summer, and October (0.25) and November (6.86) are the months after disappearing Indian summer monsoon.

Farur gazelles
Based on the observation of the DoE, the breeding season of the Farur gazelles starts in November before the beginning of boreal winter (December-March), when monthly precipitation increases (average total monthly precipitation: 28.88) and temperatures decrease (average total monthly temperature: 20.77), probably the best time for young to be born. Around 350 (range: 187-519) gazelle individuals exist on Farur at any one time. They are mainly browsers (diurnal and partially nocturnal), feeding on foliage, flowers, and seed pods of A. tortilis (Fig. 5) and other shrubs, but also graze on grasses and herbs, like other populations of G. gazella and G. arabica do [34][35][36]. Acacia trees produce a large number of pods that are eaten by gazelles. In the dry season the gazelles recently started to feed on forage provided by the DoE. The soil of the island is bare on many days of the year, and it seems that the percentage of grass cover and the height of A. tortilis are important environmental variables affecting the presence of G. a dareshurii in spring [37].

DNA extraction, amplification, and sequencing
Thirty tissue samples were collected from dead gazelles on the island after a flood event in 2011 (Table 3) and preserved in 96% ethanol in sterile 15 ml tubes.
DNA was extracted using phenol-chloroform methods [38]. For amplification of the complete coding region of the mitochondrial cyt b we used the primers L14724 and H15915 [39]. The reaction mixture was prepared in 25 μl volume, containing 1 unit of Euro Taq DNA polymerase, 10 µM Tris-HCl, 30 µM KCl, 1.5 mM MgCl 2 , 250 µM of each dNTP and 2 pmol primers (Bioneer, South Korea). The thermocycling was performed as follows: initial denaturation (180 s at 95 °C), followed by five cycle steps of 60 s at 94 °C (denaturation), 90 s at 45 °C (primer annealing) and 90 s at 72 °C (elongation), then 40 cycle steps of 60 s at 94 °C, 60 s at 50 °C and 90 s at 72 °C, and lastly, a final extension step (600 s at 72 °C) [39]. Doublestrand cycle Sanger sequencing was performed using the Big Dye Terminator Cycle Sequencing kit v.3.1 (Applied BioSystems) and electrophoresis of the purified sequencing product was carried out on an ABI PRISM 3730xl automatic sequencer.
For a phylogenetic analysis of the genus Gazella, Lerp et al. [40,41] published a new set of nuclear intron  markers. Two of the six markers (CHD2 and ZNF618) were selected for the amplification using the primers from Lerp et al. [41] as they showed a good distinction between G. gazella and G. arabica.

Phylogenetic analyses Cyt b
A mitochondrial cyt b sequence alignment was constructed including ten new sequences from Farur gazelles (Acc. No: OL347679-OL347688) and 45 sequences already published in GenBank (Acc. No.: see Fig. 2) covering the two species of mountain gazelles Gazella arabica and G. gazella, and G. dorcas (as closest relative to the ingroup for rooting the tree). This alignment was used for the cyt b phylogenetic reconstruction. A Bayesian analysis was performed in BEAST MC 3 v.1.7.5 [42]. jModelTest v.2.1.1 [43] identified HKY + Γ as the best fitting substitution model. We used molecular clock data estimates inferred for G. dorcas [39] and ran MC 3 simulations with 10 7 generations, discarding the first 10% of the runs as burn-in.

Concatenated analysis of cyt b and two nuclear introns
Based on the alignment by Lerp et al. [41] we created a concatenated alignment of all three markers, i.e., cyt b and two nuclear introns (ZNF618 and CHD2), adding for each marker the single haplotype of the Farur gazelles to the sequences provided by Lerp et al. [41]. Sequences were aligned using the Clustal W algorithm [44] implemented in Mega v.5 [45], and final adjustments were made by eye. The final alignment has 2521 bp. The best-fitting partitioning scheme and nucleotide substitution models were estimated using greedy search algorithm with PhyML [46] in PartitionFinder v.2.1.1 [47,48]. We tested among partitioning schemes including division of protein-coding genes of cyt b into 1st, 2nd, and 3rd codon positions and two nuclear intron partitions. Models were selected by the Bayesian information criterion (BIC). We found the optimal partitioning scheme includes four partitions (optimal models are indicated in brackets) 1st codon of cyt b (K80 + I), 2nd codon of cyt b and CHD2 (HKY + I), 3rd codon of cyt b (HKY + I), and 4th ZNF618 (HKY + Γ). Bayesian inference analyses were carried out in MrBayes v.3.2 [49] with two independent runs of four Markov chains (one cold and three heated) over 10,000,000 generations and sampling every 1000 generations. The first 25% of the sampled trees and estimated parameters were discarded as burn-in. Convergence of the model parameters was monitored using the program Tracer v.1.7.1 [50]. The consensus phylogenetic tree was then edited in FigTree v.1.4.4 (http:// tree. bio. ed. ac. uk/ softw are/ figtr ee/).

Morphometric analysis
In the present study, 14 skulls of gazelles from Farur Island (8 males, 6 females) were measured, including 13 skulls from the Museum of Hormozgan Department of Environment (HDoE) and one from the Isfahan University of Technology (IUT). The 13 skulls from HDoE were collected during fieldwork by HDoE in 2011. The skull in IUT belongs to the type series that was collected after a drought in 1986 by B. Farahang Dareshuri.
Up to 50 measurements per skull were taken by D.F. based on the method described in Bärmann et al. [14]. This data set is complemented with data from Bärmann et al. [14] and Wronski et al. [16], including 22 G. a. arabica (16 males, 6 females), 7 "G. a. erlangeri" (5 males, 2 females), 6 G. a. muscatensis (3 males, 3 females), 9 "G. a. farasani" (6 males, 3 females), and 11 G. gazella (9 males, 1 female) specimens (Table 4). Missing measurements due to incomplete skulls were replaced with average values of the other specimens belonging to the same taxon and sex. Nineteen measurements (Table 5) were included in the final analyses. All values were log10-transformed as recommended by Keene [51]. The data were explored using PCA and DFA, with cross-validation to test for the distinctness of G. a. dareshurii from other G. arabica subspecies (Fig. 4, Table 2). All morphometric analyses were conducted with SPSS v.24.