Genetic structure and conservation status of Astragalus subrecognitus (Fabaceae): a very rare and narrow endemic species

1Department of Plant and Animal Biology, Faculty of Biological Science and Technology, University of Isfahan, Hezarjarib Street, P.O. Box 81746-73441, Isfahan, Iran 2Botany Research Division, Research Institute of Forests and Rangelands, Agricultural Research, Education and Extension Organization (AREEO), P.O. Box 13185-116, Tehran, Iran 3Research Department of Natural Resources, Isfahan Agricultural and Natural Resources Research Center, Agricultural Research, Education and Extension Organization (AREEO), P.O. Box 81785-199, Isfahan, Iran 4Leibniz Institute of Plant Genetics and Crop Plant Research (IPK), Corrensstrasse 3, D-06466 Seeland-Gatersleben, Germany 5German Centre of Integrative Biodiversity Research (iDiv) Halle-Jena-Leipzig, D-04103 Leipzig, Germany *Corresponding author: a.bagheri@sci.ui.ac.ir REGULAR PAPER


INTRODUCTION
Astragalus L. (Fabaceae) is with 2500-3000 species and more than 250 sections the largest genus of flowering plants in the World (Lock & Simpson 1991;Podlech & Zarre 2013). The species mostly occur in semiarid and arid mountainous areas of South America and the Northern Hemisphere. Astragalus is particularly diverse in southwestern Asia where about 1000-1500 species occur (Maassoumi 1998). Section Hymenostegis Bunge, one of the spiny sections within Astragalus, has its centre of distribution in Iran, particularly in the northwestern parts of the country (Bagheri et al. 2014). Local endemism plays an important role in the biology of this section (Zarre & Podlech 1996;Bagheri et al. 2017) and nearly all of its members have very narrow distribution areas. Narrow endemism combined with small effective population sizes might result in depletion of genetic diversity in such species, restricting their ability to cope with changing environmental conditions.
One of the endemic sect. Hymenostegis species is Astragalus subrecognitus Bagheri, Maassoumi & F.Ghahrem., a small perennial shrub (Bagheri et al. 2011). Astragalus subrecognitus grows very slowly and is a weak competitor, although by forming cushions it thrives well in wind-blown areas, thus avoiding competition with many other plants. The species seems not adapted to specific soil types but is overall rare in more rocky areas. It is entomophilous and seems to be pollinated mostly by bees. The inflorescence is usually elongated reaching above the cushion surface. When seeds are mature, the pods easily separate from the inflorescences. Given these structures and also based on our field observations, it seems that seed dispersal of this species is mainly by wind. Astragalus subrecognitus propagates most probably only through seeds, as we did not find signs of vegetative reproduction in this species. It occurs with few individuals in a restricted part of the Iranian Zanjan province. Its habitats are mountain steppes at elevations of 2100-2800 m in the vicinity of the Qezel Ozan River. As the river and bottom of the valley is 900-1600 m below the A. subrecognitus populations, and the valley floor is densely populated and used for agriculture, it effectively subdivides the species' distribution area in a western and eastern population.
Studies of genetic diversity within Astragalus species are still rare. For example, Anand et al. (2010) found two distinct genotype groups in Indian A. rhizanthus Royle ex Benth. using a set of anonymous molecular markers. Morris et al. (2016) developed and characterized microsatellites in the endangered A. bibullatus Barneby & Bridges, while inter-simple sequence repeats (ISSR) was used to analyse the narrow endemics A. oniciformis Barneby (Alexander et al. 2004) and A. crassicarpus Nutt. (Rogenski et al. 2009). Due to the few results regarding genetic diversity in endemic Astragalus species it is currently not possible to generalize results and to draw conclusions for conservation measurements. This makes additional studies of genetic diversity within Astragalus species necessary particularly for the endemic species.
To analyse the genetic variability of A. subrecognitus we are employing ISSR and sequence-related amplified polymorphism (SRAP) markers. ISSRs are microsatellite-based markers that employ the ubiquity of microsatellite repeats in the eukaryotic genomes to PCR-amplify stretches of DNA in-between microsatellite loci. This marker class has been widely used in population studies because they are variable, reliable and fast compared to other molecular markers (Wolfe & Liston 1998;Harris 1999;Esselman et al. 1999;Camacho & Liston 2001;McGlaughlin et al. 2002;Smith & Bateman   2002). SRAP, in contrast, aims for the amplification of coding regions of DNA with primers targeting ubiquitous motifs of open reading frames (Li & Quiros 2001). These markers have proven to be robust and highly variable (Robarts & Wolfe 2014). While ISSR loci are distributed throughout the entire genome, SRAP amplifies fewer and functional regions. Thus, ISSR markers might be more informative than SRAP in closely related taxa. We here use these dominant markers to infer the amount and distribution of genetic structure within A. subrecognitus, i.e. we want to evaluate if the river valley functions as a barrier to gene flow between the subpopulations resulting in even smaller effective population size for the species.

Taxon sampling
We regularly monitored the occurrence of A. subrecognitus individuals during the years 2008-2017. Few young individuals were found within the population, which indicates on going natural regeneration, however the number of seedlings is low compared to mature individuals and seedlings are not present at all stands. The area where the species thrives is its type locality and some neighbouring regions from both sides of the Qezel Ozan River in Zanjan province, Iran ( fig. 1). Leaf material was collected from a single individual at each of the 12 known stands of the species. Individual numbers at these stands range from one to three specimens but mostly only one individual is present. Herbarium vouchers were taken by removing a flowering twig without destroying the plants. They were deposited in the herbarium TARI. Sample information is given in table 1.

DNA extraction and amplification
Total genomic DNA of individuals of A. subrecognitus was extracted from leaf tissue using the method of Gawel & Jarret (1991). For performing ISSR analysis, 40 ISSR primers (Blair et al. 1999) were tested, from which eight (supplementary file 1) amplified reliably detectable and polymorphic DNA fragments from genomic DNA of A. subrecognitus accessions. The PCR amplification for ISSR was performed in 15 µL volume with 250 nM of primer, 0.2 mM of each dNTP, 1.5 mM MgCl 2 , 1 U Taq polymerase (Ampliqon), and 50-100 ng of genomic DNA. The PCR was carried out at 95°C for 4 min for initial denaturation, 40 cycles of 1 min denaturation at the primer-specific annealing temperature (supplementary file 1), ending with an extension step of 2 min at 72°C, followed by a final extension step of 10 min at 72°C. PCR products were screened on 2% agarose gels stained with ethidium bromide under UV light. For SRAP we used nine primer combinations (supplementary file 2) designed by Li & Quiros (2001). The PCRs for SRAP were performed in 25 μL reaction volumes containing Taq 2× Master Mix Red (Ampliqon), 0.1 μM of each forward and reverse primer (supplementary file 2), 50 ng DNA template and 20 μL nuclease-free water. The PCR program conducted with the following cycle profile: 5 min of initial denaturation at 94°C followed by 5 cycles of 1 min denaturing at 94°C, 1 min annealing at 35°C and 1 min of extension at 72°C. After these, 35 cycles of 1 min denaturing at 94°C and 1 min annealing at 50°C ending with an extension step of 5 min at 72°C, followed by a final extension step of 10 min at 72°C. PCR products were screened as before.

Data analysis
For both marker classes the presence (1) or absence (0) of DNA fragments amplified in PCR was scored and both datasets were combined in a single binary data matrix. Genetic similarities were calculated based on Nei and Li distances (Nei & Li 1979) in PAUP* version 4.0a163 (Swofford 2002). The Neighbor-Joining (NJ) algorithm was used in PAUP for clustering and to calculate a phenogram that shows relation- ships between the studied individuals. Branch support was evaluated by 1000 bootstrap re-samples using identical settings as before. The correlation between genetic distances and geographic distances (r) was measured using the Mantel test (Mantel 1967) implemented in GenAlEx version 6.5 (Peakall & Smouse 2006). Principal coordinates analysis (PCoA) was performed using GenAlEx. To assess genetic variability detected in both datasets, basic parameters including Nei's gene diversity index (h), Shannon index (I) and percentage of polymorphic loci (PPL) were calculated from the data using Popgene version 1.32 (Yeh et al. 2000).

RESULTS
A total of 260 bands for the combined ISSR and SRAP dataset were obtained (125 bands for ISSR, 135 bands from SRAP). The number of bands per ISSR marker ranged from 11 to 25 with an average of 18 bands per primer. The highest and the lowest numbers of bands for the ISSR markers were produced from the primer ISSR 6 (25 bands) and ISSR 880 (11 bands), respectively (supplementary file 1). The number of bands per primer combination for SRAP ranged from 6 to 26 with an average of 16 bands per primer combination. The highest and the lowest numbers of bands were detected by the primer combinations Me2+Em3 (26 bands) and Me3+Em2 (6 bands), respectively (supplementary file 2). A comparison of genetic information detected by ISSR and SRAP is provided in supplementary file 3.
The NJ analysis of genetic distances resulted in two clades separating individuals 1-8 (clade 1) from 9-12 (clade 2; fig. 2A). These genetic groups do however not concur with the geographic origins of the individuals east and west of the Qezel Ozan River, as individual 11 of clade 2 was collected in the east while 9, 10 and 12 originated from the west of the valley. This clustering is also reflected in the PCoA plot ( fig.  2) where the two main groups were separated along axis 1. PCoA axis 2 separates in addition individuals from the westernmost stands west of the river (1, 4, 5) from the remaining individuals of clade 1 (2, 3, 6, 7). In the NJ tree these westerly individuals form a monophyletic group within clade 1 and are deeply embedded within the group of individuals derived from the slopes east of the river. Individual 8 is at the basal position within clade 1 and possesses some bands otherwise occurring in the individuals of clade 2. This intermediate position is also visible in the PCoA plot and is interpreted by us as the result of hybridization between members of both main clades.
In the Mantel tests for correlation of genetic with geographic distances, the ISSR data revealed low correlation (r = 0.274, p = 0.030), while SRAP data found no correlation between genetic distance and geographic distance (r = 0.111, p = 0.180). For the combined dataset the corresponding values were r = 0.274 and p = 0.010 (supplementary file 4).

DISCUSSION
Based on our data, ISSR resulted in slightly more polymorphisms (97.60%) than SRAP (93.33%; supplementary file 3). Anand et al. (2010) also indicated high polymorphism (93.07%) of ISSR bands across the different genotypes of A. rhizhanthus. Although in our case the overall number of fragments is higher in the SRAP compared to the ISSR dataset. With the combined dataset it was possible to discern all individuals ( fig. 2A). Although our samples covered all sites of the species' occurrence ( fig. 1), at each location just one individual was collected as individual numbers are low (1-3) per site. Moreover, in a broad sense, samples from each side of the river could be considered as separate populations. The low number of studied individuals could cause a problem regarding detection of population structure but here evaluation of genetic diversity is in conflict with conservation of the species, which made us sample with lowest impact on the study sites. This is not an unusual issue among rare and narrow endemic taxa. For example, Gaafar et al. (2014) used 5 populations to represent Breonadia salicina (Vahl) Hepper & J.R.I.Wood, Ma et al. (2015) used 16 samples to study the genetic diversity of Cotoneaster schantungensis G.Klotz, Turchetto et al. (2016) used 13 populations for the rare species Petunia secreta Stehmann & Semir, and Cánovas et al. (2015) used seven populations of the endemic species Viola cazorlensis Gand. In our case, where we included 12 specimens of A. subrecognitus, we sampled all occurrence sites and about 50% of the known individuals of the species, which makes our results probably still representative.
We assume that current gene flow between the westerly and easterly populations is rather unlikely given the horizontal distance of c. 70 km in between and lack of populations low in the valley. However, during colder climate cycles in the past as, e.g., 20 000 years ago, the species' distribution might have been at lower elevations so that the river valley could be bridged by A. subrecognitus. Afterwards, with climate warming the population climbed the slopes on both sites of the valley and became isolated (again). This hypothesis of secondary contact of formerly subdivided and already differentiated populations of A. subrecognitus can explain the occurrence of members of the major genetic clusters within the extant eastern and western subpopulations. Although the recognition of admixture can be partly masked by the dominant nature of the markers we used, gene flow between the genetic clusters we found seems to be a slow process, as only one individual showed clear indications of hybridization (8) Yang et al. (2017), in a study of population differentiation of Lotus sessilifolius L. using microsatellite markers, revealed low gene flow levels between adjacent populations. Thus, the values detected in A. subrecognitus seem to be no exception within the family. Our results mean on the one hand that the populations of A. subrecognitus might be able to retain their genetic diversity for the time being and that genetic drift might play currently no major role in this species, despite the few existing individuals of the species. On the other side, the small effective population size is further enhanced by the subdivision of the population by the river valley and the absence of extant gene flow between subpopulations. From this, it would be interesting to find out how old the individuals can become and what the actual reproduction rate within these populations is. Such demographic data would be of high additional importance to finally infer the possible future fate of species with very small population sizes.

Preliminary conservation status
Astragalus subrecognitus ( fig. 3) only exists at two localities with c. 70 km maximum distance in northwestern Iran. The known area of occupancy (AOO) is only 28 km 2 , and the extent of occurrence (EOO) is 108 km 2 . There are no protection strategies for the region of the species in effect. Furthermore, there is pressure on habitats in this region caused by overgrazing that leads to high risk of habitat loss and fragmentation for A. subrecognitus. Regarding the low number of individuals of A. subrecognitus, we infer that conservation measurements for this endangered and narrow endemic species might be crucial. Hence, following the IUCN Red List criteria (IUCN 2012) A. subrecognitus is categorized as 'Endangered' (EN: B1ab(iii)+2ab(iii)).

ACKNOWLEDGMENTS
This study was performed with the financial support of the University of Isfahan. We wish to thank the authorities of the herbaria TARI and HUI for providing herbarium facilities for our study. Also, we are grateful to the two anonymous reviewers and Dr. Katharina Budde for their valuable comments on the manuscript.