Identification of the Efficacy of Ex Situ Conservation of Ammopiptanthus nanus Based on Its ETS-SSR Markers

Ammopiptanthus possesses ancestral traits and, as a tertiary relict, is one of the surviving remnants of the ancient Mediterranean retreat and climate drought. It is also the only genus of super xerophytic, evergreen, broad-leaved shrubs. Ammopiptanthus nanus, one of the two species in this genus, is predominantly found in extremely arid and frigid environments, and is increasingly threatened with extinction. Study of the species’ genetic diversity is thus beneficial for its survival and the efficacy of ex situ conservation efforts. Based on transcriptome data, 15 pairs of effective EST-SSR were screened to evaluate A. nanus genetic diversity. In all, 87 samples from three populations were evaluated, the results of which show that ex situ conservation in the Wuqia region needs to be supplemented. Conservation and breeding of individual A. nanus offspring should be strengthened in the future to ensure their progeny continue to exhibit high genetic diversity and variation.


Introduction
Ammopiptanthus, an oligotypic legume and Tertiary relict plant of the ancient Mediterranean, is the only super xerophytic, evergreen, broad-leaved shrub endemic to desert areas in central Asia [1]. It includes two species, namely, Ammopiptanthus nanus (M. Pop.) Cheng f. and Ammopiptanthus mongolicus (Maxim. ex Kom.) Cheng f., both of which are listed in the China Plant Red Data Book as a Rare and Endangered Plant, and also listed as critically endangered in the IUCN Red List Category and Criteria [2,3]. The enormous root structure of A. nanus is physiologically and ecologically adapted to difficult environments, allowing it to live on barren mountains and the stony Gobi under significant drought conditions by preventing wind and fixing sand [4]. Temperatures in the distribution area of A. nanus frequently reach over 35 • C in the summer and −20-30 • C or lower in the winter. The process of soil humus accumulation is weak in the region, and the soil has low fertility; therefore, the sand holly has acquired traits such as early resistance, cold resistance, infertility resistance, and a developed root system [5]. According to the experience of herdsmen, running the branches of A. nanus through a camel's nose can prevent infection [6]. Additionally, the shrub's broad-leaved, evergreen character in desert areas as well as its thick leaves, golden yellow corolla, and short plant body lend it utility as an ornamental flower in the spring [7]. As the raw material for extracting industrial oil containing various special alkaloids and flavonoids in its stems, branches, leaves, and seeds, A. nanus has great value for research and development in the biological industries [8]. Various quinolizidines including lupanine, sparteine, and their derivatives have been isolated from the leaves of A. nanus [9]. A. nanus is a relatively primitive legume species belonging to the third Plants 2023, 12, 2670 3 of 20 SSR consists of short (generally 1-6 bp) tandem replication of nucleotide motifs and is characterized by a high degree of variation, codominant inheritance, and wide distribution. SSR is recognized as one of the best molecular markers [28], and its simple operation is a common means of analyzing molecular genetic diversity [29]. SSR markers are superior to SNP and AFLP markers for detecting polymorphism because SSR is an effective multipleallele marker, whereas SNP and AFLP are single-allele markers [30]. The features and co-dominance of SSR markers make them the preferred DNA markers for assessing plant genetic diversity [31]. In particular, the highly conserved nature of EST-SSR makes it more accurate than G-SSR at predicting genetic differences and more generalizable across similar species [32,33].
In this study, we utilized EST-SSR markers developed from the transcriptome genes of A. nanus to assess the efficacy of ex situ conservation efforts. By investigating the differences in genetic diversity between the original population in Wuqia and an ex situ population at the molecular level, we were able to identify the impact of ex situ conservation on A. nanus. Thus, the study provides a theoretical foundation for the ex situ conservation, breeding, and inheritance of A. nanus.

Library Quality Management, Data Sequencing and Quality Assurance
Following the preceding series of analyses and quality checks, we obtained 14.57 GB of clean data, and the proportion of Q30 bases in each sample did not fall below 92.77%. Two samples of clean data are presented in Table 1. 1,2 Samples: sample name; Read Number: total number of pair-end Reads in clean data; Base Number: total number of bases in clean data; GC Content: clean data GC content, which refers to the ratio of G and C bases to the total number of bases in the clean data; % ≥ Q30: percentage of bases with clean data quality value greater than or equal to 30.

Transcriptome Sequencing Library Data Assembling
Qualified transcriptome sequencing libraries are essential for reliable transcriptome data analysis results [34]; the clean data from the two samples were compared with the assembled Transcript or Unigene libraries [35]. The statistical results of this comparison are presented in Table 2.

Transcriptome Sequencing Library Quality Assessment
To assure the quality of transcriptome sequencing libraries, these were evaluated from three distinct vantage points. The randomness of mRNA fragmentation and mRNA degradation were assessed by examining the distribution of insert fragments on Unigene. After mRNA fragmentation, the insert fragment size selection can be interpreted as the random extraction of subsequences from mRNA sequences. If the sample size (number of mRNAs) is larger, the interruption method and time control are more suitable, and the probability of each part of the target RNA being extracted is closer. That is, the randomness of mRNA fragmentation is greater, and the reads coverage on mRNA is more uniform. The degree of randomness of mRNA fragmentation was examined by simulating the results of mRNA fragmentation through the distribution of mapped reads in Unigene. Each sample's positional distribution of mapped reads on mRNA transcripts was obtained ( Figure 1).
After mRNA fragmentation, the insert fragment si e selection can be interpreted as random extraction of subsequences from mRNA sequences. If the sample si e (numb mRNAs) is larger, the interruption method and time control are more suitable, and probability of each part of the target RNA being extracted is closer. That is, the rand ness of mRNA fragmentation is greater, and the reads coverage on mRNA is more form. The degree of randomness of mRNA fragmentation was examined by simula the results of mRNA fragmentation through the distribution of mapped reads in Unig Each sample's positional distribution of mapped reads on mRNA transcripts was tained ( Figure 1). The hori ontal axis in Figure 1 represents the mRNA position, while the vertical represents the percentage of reads in mapped reads within the corresponding posi interval. Because the lengths of the reference mRNA varied, each mRNA was divided 100 intervals based on its length, and the number and proportion of mapped reads in e interval was determined. During library preparation, the dispersion of insert length flects the effect of gel cutting or magnetic bead purification ( Figure 2). The hori ontal represents the distance between the starting and ending points of the double-ended re The horizontal axis in Figure 1 represents the mRNA position, while the vertical axis represents the percentage of reads in mapped reads within the corresponding position interval. Because the lengths of the reference mRNA varied, each mRNA was divided into 100 intervals based on its length, and the number and proportion of mapped reads in each interval was determined. During library preparation, the dispersion of insert lengths reflects the effect of gel cutting or magnetic bead purification ( Figure 2). The horizontal axis represents the distance between the starting and ending points of the double-ended reads in the Unigene library, ranging from 0 to 800 bp, and the vertical axis represents the number of double-ended reads or insert fragments at various distances between the starting and ending points of the comparison.
Finally, the library capacity and the sufficiency of reads (mapped reads) relative to the Unigene library was assessed by plotting the saturation. Sufficient valid data are required for accurate analysis. There is a positive correlation between the number of genes detected by transcriptome sequencing and the amount of sequencing data; the greater the quantity of sequencing data, the greater the number of genes detected. However, the number of genes in a species is limited, and gene transcription is time-and space-specific, so the number of detected genes tends to saturate as the sequencing quantity increases. To determine if the data are sufficient, it is necessary to determine if the number of newly detected genes decreases with increasing quantity of sequencing data and signaling saturation. Using mapped reads for each sample, the saturation of the number of detected genes was simulated and depicted on a graph ( Figure 3).
Plants 2023, 12, x FOR PEER REVIEW 5 o in the Unigene library, ranging from 0 to 800 bp, and the vertical axis represents the nu ber of double-ended reads or insert fragments at various distances between the start and ending points of the comparison. Finally, the library capacity and the sufficiency of reads (mapped reads) relative the Unigene library was assessed by plotting the saturation. Sufficient valid data are quired for accurate analysis. There is a positive correlation between the number of ge detected by transcriptome sequencing and the amount of sequencing data; the greater quantity of sequencing data, the greater the number of genes detected. However, the nu ber of genes in a species is limited, and gene transcription is time-and space-specific the number of detected genes tends to saturate as the sequencing quantity increases. determine if the data are sufficient, it is necessary to determine if the number of new detected genes decreases with increasing quantity of sequencing data and signaling sa ration. Using mapped reads for each sample, the saturation of the number of detec genes was simulated and depicted on a graph ( Figure 3). Saturation curves were generated by dividing mapped reads into 100 equal portions and progressively increasing the number of genes detected by viewing the data. The horizontal axis represents the number of reads (in 10 6 units), while the vertical axis represents the number of genes detected (in 10 3 units). If the slope gradually flattens as the curve extends to the right, it indicates that as the amount of sequencing data increases and tends to saturate with an adequate amount of valid sequencing data, each sample contains fewer and fewer newly detected genes.

Annotation of Unigene Function
Unigene Orthology results were obtained using the BLAST [36] software to compare Unigene sequences with the NR [37], Swiss-Prot [38], GO [39], COG [40], KOG [41], eggNOG4.5 [42], and KEGG [43] databases, and by obtaining KEGG in KEGG using KOBAS2.0 [44]. After predicting the amino acid sequence of Unigene, the software HM-  Saturation curves were generated by dividing mapped reads into 100 equal portio and progressively increasing the number of genes detected by viewing the data. The h i ontal axis represents the number of reads (in 10 6 units), while the vertical axis represe the number of genes detected (in 10 3 units). If the slope gradually flattens as the cur extends to the right, it indicates that as the amount of sequencing data increases and ten to saturate with an adequate amount of valid sequencing data, each sample contains few and fewer newly detected genes.

Annotation of Unigene Function
Unigene Orthology results were obtained using the BLAST [36] software to comp NCBI's NR database is a non-redundant protein database that contains the Swiss-Prot, PIR (Protein Information Resource), PRF (Protein Research Foundation), and PDB (Protein Data Bank) protein databases, as well as translated CDS data from GenBank and RefSeq. The EBI (European Bioinformatics Institute) maintains the Swiss-Prot database, which contains a highly reliable database of annotated protein information with relevant references and revision. The GO (Gene Ontology) database is a standardized international gene function classification system that offers a dynamically updated standard vocabulary for describing the functional properties of genes and gene products in an organism. The Molecular Function, Cellular Component, and Biological Process categories of the Gene Ontology (GO) database each describe the molecular functions that a gene product may execute, as well as the cellular environment and biological processes in which it is involved. The COG (Clusters of Orthologous Groups) database is a homology classification database for gene products and an early database for identifying direct homologous genes, based on extensive comparisons of protein sequences from numerous organisms. The KOG (euKaryotic Orthologous Groups) database is intended for eukaryotes and classifies homologous genes from various species into different orthologous clusters based on direct homology of genes combined with evolutionary relationships; KOG has 4852 classifications at present. Genes from the same orthologous clusters have the same function, allowing functional annotations to be directly inherited by other KOG cluster members. The eggNOG (v4.5) database contains functional descriptions and functional classifications of immediate homologous proteins, combining COG, KOG, and many more proteins; it covers significantly more protein sequences than COG and KOG, and it includes 5228 viral proteins. The KEGG (Kyoto Encyclopedia of Genes and Genomes) database is a database that comprehensively analyzes the metabolic pathways and functions of gene products in cells. It includes metabolic pathways (PATHWAY), pharmaceuticals (DRUG), diseases (DISEASE), gene sequences (GENES), and genomes (GENOME). Use of this database facilitates the investigation of gene and expression data as a holistic network.
The Pfam (Protein family) database is the most comprehensive classification system for annotating protein structural domains by constructing HMM statistical models from the amino acid sequences of each family via protein sequence alignment. Proteins are composed of one or more structural domains, which are functional regions, and exhibit sequence conservation. The functions of proteins can be predicted based on the sequence of structural domains. Selecting a BLAST parameter E-value that was ≤1 × 10 −5 and an HMMER parameter E-value that was ≤10 −10 yielded a total of 39,437 Unigene sequences with annotation information for this research (Table 3).

EST-SSR the Select of Primers
A. nanus gene sequences were screened using MISA (Table 4). In a transcriptome genome of approximately 42.52 Mb, there were 11,645 SSR marker repeats with a total length of 169,920 bp. The frequency of SSR markers in the gene sequence was 273.88 SSRs/Mb, with a density of 3996.48 bp/Mb, and the transcriptome genome's total SSR content was 0.4%. Figure 4 summarizes the results, classified by length of the SSR loci. All loci ranged in length from 10 to 42 bp, with an average of 14.59 bp. Overall, 10 bp was the most common repeating unit, occurring 1945 times. It was followed by 15 bp, 12 bp, and 18 bp, with respective frequencies of 1634, 1447, and 1391.
The statistical analysis of the number of repeats of SSR loci ( Figure 5) showed that all repeats fell within the range of 5 to 24. The maximum proportion of duplicates among them was 10, accounting for 18.78% of the total. This was followed by 5, 6, and 11 times, representing 13.39%, 12.52%, and 10.57% of the duplicates, respectively. Total content of transcriptome genome SSRs (%) 0.40 Figure 4 summari es the results, classified by length of the SSR loci. All loci ranged in length from 10 to 42 bp, with an average of 14.59 bp. Overall, 10 bp was the most common repeating unit, occurring 1945 times. It was followed by 15 bp, 12 bp, and 18 bp, with respective frequencies of 1634, 1447, and 1391.  them was 10, accounting for 18.78% of the total. This was followed by 5, 6, and 11 times, representing 13.39%, 12.52%, and 10.57% of the duplicates, respectively. The repetitive unit classes of SSR loci are summari ed in  Table 6 lists the principal SSR sequence repeat units and the corresponding numbers of repeats.
The loci results for A. nanus are depicted in Figure 6, where the vertical axis represents the intensity of fluorescence, and the hori ontal axis represents the si e of SSR prod- The repetitive unit classes of SSR loci are summarized in SSR sequences contained a total of 238 nucleotide repeats; Table 6 lists the principal SSR sequence repeat units and the corresponding numbers of repeats. The loci results for A. nanus are depicted in Figure 6, where the vertical axis represents the intensity of fluorescence, and the horizontal axis represents the size of SSR product fragments. Table 7 shows that the total genetic diversity (Ht) of the 15 primer pairs detected in 87 samples ranged from 0.691 to 0.904, with a mean value of 0.83. The mean observed heterozygosity (Mean Ho) ranged from 0 to 0.55, with a mean value of 0.25. The mean expected heterozygosity (Mean He) ranged from 0.218 to 0.716, with a mean value of 0.47. The ranges of the intra-population inbreeding coefficient (Fis),total population inbreeding coefficient (Fit), and genetic differentiation coefficient (Fst) are −0.021 to 1, 0.326 to 1, and 0.208 to 0.733, respectively. Gene flow (Nm) ranged from 0.091 to 0.951, with a mean value of 0.43. Values of Nm which are <1 indicate that differentiation between populations may occur due to genetic drift, and that there is a low level of gene flow among most populations. As shown in Table 7, F6 is the most efficient primer, followed by F3 and F1. F2 produces fewer effective results. Plants 2023, 12, x FOR PEER REVIEW 11 of 23 Figure 6. Partial genotyping peak plots of polymorphic SSR primers for the representative samples of A. nanus. Table 7 shows that the total genetic diversity (Ht) of the 15 primer pairs detected in 87 samples ranged from 0.691 to 0.904, with a mean value of 0.83. The mean observed hetero ygosity (Mean Ho) ranged from 0 to 0.55, with a mean value of 0.25. The mean expected hetero ygosity (Mean He) ranged from 0.218 to 0.716, with a mean value of 0.47. The ranges of the intra-population inbreeding coefficient (Fis),total population inbreeding coefficient (Fit), and genetic differentiation coefficient (Fst) are −0.021 to 1, 0.326 to 1, and 0.208 to 0.733, respectively. Gene flow (Nm) ranged from 0.091 to 0.951, with a mean value of 0.43. Values of Nm which are <1 indicate that differentiation between populations may occur due to genetic drift, and that there is a low level of gene flow among most populations. As shown in Table 7, F6 is the most efficient primer, followed by F3 and F1. F2 produces fewer effective results.  At the populations' level, 15 pairs of SSR primers revealed the population genetic diversity of a total of 87 samples from the three populations (Table 8). The number of alleles (Na) ranged from 3.6 to 6.333, containing a total of 14.93, with a mean value of 4.978; the effective number of alleles (Ne) ranged from 2.429 to 2.714, with a mean value of 2.58; the Shannon's information index (I) ranged from 0.827 to 1.121, with a mean value of 0.969; the observed heterozygosity (Ho) ranged from 0.231 to 0.29, with a mean value of 0.254; expected heterozygosity (He) ranged from 0.409 to 0.535 with a mean value of 0.466; genetic diversity parameters of the populations (PPB) ranged from 0.7333 to 0.9333, with a mean value of 82.22% and the highest, at 93.33%, in the Wuqia region. This indicates that the top priority for conservation of A. nanus genetic diversity is in the Wuqia region. The genetic distance and genetic differentiation coefficient (FST) are important parameters for evaluating the degree of genetic difference and similarity among A. nanus populations. These were calculated among different populations of A. nanus based on Nei's genetic distances using the GenAlEx software ( Table 9). The genetic distances between populations of A. nanus ranged from 2.659 to 4.9, with the largest genetic distances between the Wuqia and Kyrgyzstan populations and the smallest between the ex situ and Kyrgyzstan populations. The genetic differentiation coefficients between populations ranged from 0.352 to 0.431, with the largest between the Wuqia and Kyrgyzstan populations and the smallest between the Wuqia and Kyrgyzstan populations. The genetic differentiation coefficient between populations ranged from 0.352 to 0.431, with the Wuqia and Kyrgyzstan populations having the largest genetic differentiation coefficient and the Wuqia and ex situ populations having the smallest genetic distance. Molecular ANOVA of the genetic variation within and between populations of A. nanus, as shown in Table 10, indicates that 29% of the genetic variation in the 87 A. nanus germplasms collected from the three populations originated between populations, while nearly 71% of the genetic variation originated within populations (Figure 7). The population genetic structure was analyzed using Structure Harvester, and the mean value corresponding to each K value was calculated ( Figure 8). The maximal inflection point occurred when K = 3, indicating that the population genetic structure had a theoretical population size of 3. Visual coordinates and primary coordinate analysis were utilized to depict data similarity or dissimilarity for study sample populations. The data from 87 samples were displayed and processed to create PCoA scatter plots (Figure 9).   Visual coordinates and primary coordinate analysis were utili ed to depict data similarity or dissimilarity for study sample populations. The data from 87 samples were displayed and processed to create PCoA scatter plots (Figure 9). The results show that the three populations of A. nanus still belong to the same species but are extremely differentiated due to different natural conditions. There is substan- Visual coordinates and primary coordinate analysis were utili ed to depict data similarity or dissimilarity for study sample populations. The data from 87 samples were displayed and processed to create PCoA scatter plots (Figure 9). The results show that the three populations of A. nanus still belong to the same species but are extremely differentiated due to different natural conditions. There is substantial genetic variation both within and between populations, and within-population variation is greater. The high degree of genetic divergence across populations may be the result of genetic drift or the combined influence of natural selection and genetic differentiation induced by gene flow between places.  Table 11. The results show that the three populations of A. nanus still belong to the same species but are extremely differentiated due to different natural conditions. There is substantial genetic variation both within and between populations, and within-population variation is greater. The high degree of genetic divergence across populations may be the result of genetic drift or the combined influence of natural selection and genetic differentiation induced by gene flow between places.

Discussion
Ex situ conservation conserves a species germplasm through artificial breeding in order to increase population sizes and prevent extinction in the wild [47]. There has been extensive work conducted in this area through methods like artificial seeding and propagation in botanical gardens and natural environments. However, these efforts can be expensive, and their efficacy has not been confirmed [48]. Inbreeding can lead to decreased fitness and thus endanger the survival of small populations [49] through what is known as inbreeding depression [50]. Genetically, this phenomenon may occur because favorable alleles at one locus are typically dominant, allowing rare and recessive deleterious alleles at the same locus to persist in a population [51]. The increased purity caused by inbreeding increases the likelihood of deleterious alleles occurring as pure heterozygotes. When this occurs, the effects of deleterious alleles cannot be concealed by dominant favorable alleles, resulting in inbreeding depression [52]. It appears that dominance is the most prevalent cause of inbreeding depression [53]. This has the potential to accelerate the extinction of A. nanus [54].
Due to the present scarcity of A. nanus populations, it is crucial to avoid inbreeding. Purifying deleterious alleles is one effective method for avoiding inbreeding depression [55]. Inbreeding increases the purity of alleles and, consequently, the likelihood that the associated deleterious traits will manifest. This may result in the elimination of deleterious alleles from congenic populations [56]. However, in extremely tiny populations, the effect of removal is also quite limited [57]. There are few direct indications from natural populations that purification can eradicate inbreeding depression [58]. Studies have shown that deleterious alleles are eliminated over time [59]. In the past, it has been proposed to eliminate deleterious alleles from small populations through artificial inbreeding. Nevertheless, many biologists believe the associated hazards outweigh the potential benefits [60]. SSR markers are valuable instruments for analyzing genetic diversity and genetic structure due to their high information density, homology, and multiple alleles [61]. Our predecessors conducted a genomic survey and SSR analysis of A. nanus homologs of A. mongolicus [62]. Several molecular investigations of A. nanus have also been conducted, including chloroplast genome comparisons, phylogenetic analyses [63], and RAPD analyses of genetic diversity [64]. In addition, it is possible to derive numerous specific markers associated with specific resistance traits, which provides a clearer indication of the level of kinship and genetic diversity among sampled materials [65]. SSR not only makes it possible to identify and evaluate the efficacy of ex situ conservation of A. nanus, but it also provides a valuable tool for future population genetics research and conservation initiatives.
In the Wuqia region, there are only a few populations and individuals of A. nanus, but their descendants exhibit a high level of genetic diversity and progeny variation. From the results of this study, they can thus be divided into two conventional protected areas, with the Kyrgyz subspecies in its own protected area. This study indicates that A. nanus has a high level of individual genetic diversity, so both in situ and ex situ conservation should be implemented for this population along with abundant genetic variation and rare alleles. A. nanus must also be conserved in situ to assure genetic diversity, as is the case with species that are widely distributed but rapidly declining [66]. The results of this investigation indicate that the ex situ conservation of Wuqia germplasms requires strengthening. Significant investments in protected areas such as the Kyrgyz area are required to reduce the risk of extinction. However, A. nanus has a very limited range, as evidenced by its likelihood of genetic isolation, genetic drift, and absence of genetic mutations [67]. Additionally, it is common for plants to decline due to hard inbreeding, particularly for such narrowly distributed species, so the potential for genetic diversity loss is high [68]. Inbreeding may be another major threat.
SSR analysis for A. nanus alone, on the other hand, has clearly identified the distinct communities of A. nanus generated by the environment. GxE has long been regarded as a critical component in plant breeding [69]. With the advent of genomics technology, several good predictors for the genotypic dimension of GxE interactions have been presented [70,71]. The ex situ conservation did not occupy all the germplasm resources, especially lacking the genotypes in Kyz. However, we possess the fundamental theory and methodology to improve the conservational efficiency of future work using more clarified evidence on GxE of A. nanus. Due to habitat devastation in the wild, ex situ conservation is an extremely effective conservation measure for species that are critically endangered. Establishing botanical gardens is also among the most effective methods for ex situ conservation of endangered species such as A. nanus in TEBG [72]. Nonetheless, the genetic diversity of ex situ plants in botanical gardens has been largely ignored or inadequately evaluated [73].
Given the increasing threats to genetic diversity posed by inbreeding and environmental change, genetic characterization is necessary to prevent the decline of these new populations and maximize their evolutionary potential [74]. While population genetic data are rarely incorporated into species assessments and management plans, this does not mean that conservationists are unwilling to incorporate genetics into their practices [75]. This suggests that, while researchers are most likely to conduct conservation genetics research, they are often unable to link policy and conservation outcomes in practice [76,77]. With more frequent collaboration between researchers and governments, there may be a larger role for genetics in future conservation programs. In this research, 87 samples of A. nanus from three populations were analyzed using ETS-SSR molecular markers to evaluate the efficacy of ex situ conservation and support further conservation and artificial breeding of this germplasm resource.
A. nanus remains on the verge of extinction. For comprehensive conservation of this germplasm resource, it is necessary to focus on the genetic diversity of ex situ plant offspring as well as the quantitative conservation of individuals. Doing so will ensure the reproductive renewal of this extremely small population. Additionally, it is important to strengthen government collaboration with botanical gardens by incorporating the results of research into resource-management policy. Effective cooperation will have a positive influence on the conservation and utilization of the unique genes of A. nanus. Future research and investment need to be increased to demonstrate the value of A. nanus for adapting to extreme environments.
Given the distributional characteristics of A. nanus, we make the following recommendations for its conservation in the future: 1.
The government should intervene to increase investment in in situ conservation and broadcast relevant policies. Seeds of different populations in A. nanus with high genetic diversity should be collected for propagation in order to obtain the maximum available genetic diversity.

2.
Plants from different populations should be introduced for appropriate crosses to increase the level of biodiversity of A. nanus. Further research on the reproductive biology of A. nanus is required to ensure the effective conservation of the species.

RNA Sample Detection and RNA Library Development
At the beginning of the study, one sample of A. nanus was collected from the Turpan Eremophytes Botanic Garden (CAS) (TEBG, 42.3 • N 89.2 • E) for transcriptome sequencing. The Nanodrop [78], Qubit 2.0 [79], and Aglient 2100 [80] methods were used to verify the purity, concentration, and integrity of the RNA samples to ensure that sufficiently high-quality samples were used for transcriptome sequencing.
After testing and qualifying the samples, the library was constructed according to the following procedures. First, eukaryotic mRNA was enriched with oligo (dT)-coated magnetic particles [81]. Fragmentation buffer was then added to interrupt the mRNA in a random manner. The first cDNA strand was then synthesized using mRNA as a template and six-base random hexamers, followed by the addition of buffer, dNTPs, RNase H, and DNA polymerase I to synthesize the second cDNA strand [82]. cDNA was purified with the aid of AMPure XP particles. Then, end repair, A-tail addition, and sequencing junction ligation were performed. The fragment size was then selected using AMPure XP beads [83]. Finally, the cDNA library was enriched via PCR.

Library Quality Management, Data Sequencing and Quality Assurance
After library construction, the library concentration and insert size were analyzed using Qubit 2.0 and Agilent 2100, respectively, and the effective library concentration was quantified using the Q-PCR method [84]. After ensuring qualification, Illumina Hiseq high-throughput sequencing was performed [82].
Based on sequencing-by-synthesis (SBS) technology, the cDNA libraries were sequenced using the Illumina HiSeq high-throughput sequencing platform, which can generate a large number of high-quality reads [85]. These reads or bases generated by the sequencing platform are referred to as raw data, and the majority of their base quality scores can meet or exceed Q30 (99.9% identification accuracy).
Before continuing with the subsequent analysis, it was necessary to ensure that the reads used were of sufficient quality to ensure the accuracy of sequence assembly and the subsequent analysis. In addition, there were very few reads containing sequencing primers, connectors, and other artificial sequences in the raw data, so it was necessary to truncate the sequencing connectors and primer sequences in the reads and filter the low-quality data to ensure data quality [86].

Transcriptome Sequencing Data Compilation
After obtaining high-quality sequencing data, the sequence was assembled. Trinity [87] is a transcriptome sequencing assembly software designed for high-throughput sequencing. The sequencing depth of a transcript is influenced by the quantity of sequencing data and other factors, as well as the abundance of the transcript's expression. The sequencing depth has a direct impact on the quality of the assembly [88].
In the beginning, the k-mer library was created by breaking the sequencing reads into shorter k-mers using the Trinity software and removing any k-mers that contained errors. Next, the k-mer with the highest frequency was chosen as the seed for greedy extension at both ends, and the process was repeated until the k-mer library was exhausted [89]. Next, the contigs were clustered to produce components, the De Bruijn graph was constructed for each component, and the nodes and edges were merged and trimmed to simplify the De Bruijn graph [90]. Finally, the De Bruijn graph [91] was deconstructed using the actual read in order to identify the transcribed sequences in each fragment collection independently.

Plant Materials in Populations
The ETS-SSR markers in plants were used to assess the efficacy of ex situ A. nanus conservation efforts. Plant samples were collected from Wuqia (50) and an ex situ reserve (24), as well as Kyrgyzstan (13), to identify genetic diversity in different regions and protect germplasm resources (Table 11).

DNA Extraction
Sampled plant materials were stored in liquid nitrogen following field collection. After transport to the laboratory, about 0.25 g fresh leaves was ground into a powder in liquid nitrogen. Next, the CTAB method was used to extract their DNA following standard procedure. The concentration was measured by spectrophotometry, diluted to about 0.10-0.25 mg/µL, and refrigerated at −20 • C for later use [90].

PCR Amplification and EST-SSR Primers Selection
The primers and DNA template described above were used for PCR reaction. The PCR reaction system was 25 µL, as follows: ddH 2 O 9.5 µL, Mix12.5 µL, primers at both ends 1 µL, and DNA1 µL. The PCR reaction process was as follows: denaturing at 94 • C for 2 min; 30 cycles at 94 • C for 1 min, annealing temperature 53-56 • C for 1 min and 72 • C for 2 min; and 72 • C for 5 min. The PCR products were subjected to 1% agarose gel electrophoresis using an electrophoresis system running the buffer solution of 1 × Tris EDTA (TE), with a constant voltage of 130 V and an electrophoresis time of 15 min. Successful amplifications (with the clear band) were sent to Shanghai SBS, Biotech Ltd. for analysis. Then, 15 pairs (Table 12) of 2-5 base SSR sequences with 3-4 repeats (about 22 bases) were selected for primer synthesis by a bio-company (Sangon, Shanghai, China) with 5 fringe HEX fluorescence labeling.

Statistical Analyses
MISA [92] is a Perl-based script that identifies SSRs from FASTA sequences and has become the tool of choice for most EST-SSR researchers due to its ease of use, lack of networking, minimal hardware requirements, and compatibility with multiple tools.
GeneMarker [93] was used to analyze the original data obtained using high-performance capillary electrophoresis (HPCE). GenAlEx [94] was used to calculate the population genetic parameters of each primer pair, including the number of alleles per locus (Na), observed heterozygosity (Ho), expected heterozygosity (He), effective number of alleles (Ne), and information index and percentage of polymorphic sites (PPB) values. Principal component analysis (PCoA) was also performed in GenAlEx. Additionally, an analysis of molecular variance (AMOVA) function was adhibited in GenAlEx to measure ex situ conservational efficacy for A. nanus within and among populations, and within and among groups.
In total, 999 random permutations were implemented in GenAlEx for each sample. The genetic structures of three (Wuqia, Kyrgyzstan, and ex situ reserve) and two (Wuqia, Kyrgyzstan, no ex situ reserve) natural populations of A. nanus were analyzed based on Bayesian analysis using the STRUCTURE software (version 2.3.4) [95]. Ten independent runs were performed for each K value. Then, results were submitted to the online software Structure Harvester to analyze the optimal K value, which was determined by the method described by the burn-in period iterations and Markov chain Monte Carlo repetitions. Finally, Core Finder was used to calculate the size of individual protected germplasm resources.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.