Development of a SNP panel for geographic assignment and population monitoring of jaguars (Panthera onca)

The jaguar (Panthera onca) is an iconic top predator that is threatened by habitat loss and fragmentation, along with an emerging expansion of poaching for the illegal trade of live individuals and their parts. To address the need for tools that improve surveillance and monitoring of its remaining populations, we have developed a genome-enabled single nucleotide polymorphism (SNP) panel targeting this species. From a dataset of 58 complete jaguar genomes, we identified and selected highly informative SNPs for geographic traceability, individual identification, kinship, and sexing. Our panel, named ‘Jag-SNP’, comprises 459 SNPs selected from an initial pool of 13,373,949 markers based on the inter-biome FST, followed by rigorous filtering and addition of eight sex-linked SNPs. We then randomly selected subsets of this panel and identified an 84-SNP set that exhibited a similar resolving power. With both the 459-SNP panel and its 84-SNP subset, samples were assigned with 98% success to their biomes of origin and 65-69% of them were assigned to within 500 km of their origin. Furthermore, ca. 10-18 SNPs within these panels were sufficient to distinguish individuals, while 6 sex-linked SNPs perfectly separated males and females. We used whole-genome data from an additional 18 jaguars to further test these panels, which correctly recovered kinship relationships and allowed inference of geographic origin of samples collected outside the spatial scope of the original sample set. These results support the strong potential of these panels as an efficient tool for application in forensic, genetic, ecological, behavioral and conservation projects targeting jaguars.

Methods
From a dataset of 58 complete jaguar genomes from Brazil (Sartor et al. [in prep]), we used a genotype likelihood approach to identify 13,373,949 SNPs, that were later filtered according to their coverage, depth, mapping quality, missing data, minor allele frequence, linkage disequilibrium, repetitive regions and Hardy-Weinberg Equilibrium. With the remaining 83k SNPs, we grouped individuals according to their biomes of origin (Amazon: n=18; Atlantic Forest: n=14; Cerrado: n=14; Caatinga: n=6; Pantanal: n=6) and calculated pairwise FST values between biomes.

To select the most informative sites for geographic assignment, we ranked the 83k SNPS based on their pairwise FST values (from highest to lowest) between the sampled biomes and kept the 50 loci with the highest FST from each pairwise comparison. We identified a panel comprising 459 SNPs. This set included the top 50 SNPs of each pairwise biome comparison; 41 SNPs were retrieved in more than one comparison and were kept only once in this selected subset. Then, we randomly selected subsets of this 459-SNP panel to assess whether smaller sets of SNPs (which can be genotyped more quicky and affordably) provide similar levels of information. As an example, we identified a set with only 84 SNPs that exhibited a similar resolving power, correctly separating individuals by biomes of origin in a PCA.

To select SNPs that were informative for sexing, we identified 698 SNPs located on the X chromosome and searched for sites that were close to the X-linked genes Amelogenin (AMELX) and Zinc-finger protein (ZFX), which have been used for sex identification of jaguars and other felids (Pilgrim et al. 2005). Subsequently, we analyzed the genotypes of these SNPs in each individual and visualized their distribution in males and females using a PCA. We selected 6 SNPs among these that showed a clear-cut pattern in which all females were homozygous and males were either heterozygous or homozygous for the alternative allele. These SNPs were located in the X-linked genes ZFX, FRMPD4, TRAPPC2, TXLNG, and USP9X.

美洲豹（Panthera onca）是标志性的顶级捕食者，正面临栖息地丧失与片段化的威胁，同时针对活体个体及其身体部位的非法偷猎活动也呈加剧趋势。为开发能够提升其残存种群监测与监管能力的工具，我们针对该物种构建了基因组辅助的单核苷酸多态性（single nucleotide polymorphism, SNP）分型面板。我们从58套完整的美洲豹基因组数据集中，筛选并鉴定出适用于地理溯源、个体识别、亲缘关系鉴定及性别鉴定的高信息价值SNP位点。本研究构建的“Jag-SNP”分型面板，初始筛选自13373949个标记位点，基于生物群系间的群体分化系数（FST）值进行筛选，经严格过滤后新增8个性连锁SNP位点，最终包含459个SNP位点。随后我们对该面板进行随机子集抽样，鉴定出包含84个SNP的子集，其分辨能力与原面板相当。利用459个SNP的完整面板及其84个SNP的子集，我们可将98%的样本成功分配至其原生生物群系，且65%至69%的样本可被分配至原产地500公里范围内。此外，上述面板中仅需约10至18个SNP位点即可完成个体区分，而6个性连锁SNP位点可完美区分雌雄个体。我们利用额外18只美洲豹的全基因组数据对上述面板进行验证，结果正确还原了亲缘关系，并可推断超出原始采样空间范围的样本的地理起源。上述结果证实，该分型面板可作为高效工具，应用于针对美洲豹的法医、遗传、生态、行为学及保护相关研究项目。

## 研究方法
本研究基于巴西地区的58套完整美洲豹基因组数据集（Sartor等，待发表），采用基因型似然法鉴定出13373949个SNP位点，随后根据覆盖度、测序深度、比对质量、缺失数据、次要等位基因频率、连锁不平衡、重复区域及哈迪-温伯格平衡进行过滤。利用剩余的83000个SNP位点，我们根据个体的原生生物群系进行分组（亚马孙：18例；大西洋森林：14例；塞拉多：14例；卡廷加：6例；潘塔纳尔：6例），并计算两两生物群系间的FST值。

为筛选适用于地理溯源的最具信息价值的位点，我们基于各采样生物群系间的两两FST值（从高到低）对83000个SNP位点进行排序，保留每一组两两比对中FST值最高的50个位点，最终得到包含459个SNP位点的分型面板。该集合包含每一组两两生物群系比对中排名前50的SNP位点，其中41个SNP位点在多组比对中被重复检出，因此在该筛选子集内仅保留一次。随后我们对该459个SNP的分型面板进行随机子集抽样，以评估更小的SNP位点集（可更快速、低成本地完成分型）是否能达到相近的信息水平。最终我们筛选出仅含84个SNP的位点集，其分辨能力与原面板相当，在主成分分析（Principal Component Analysis, PCA）中可正确根据原生生物群系区分个体。

为筛选可用于性别鉴定的SNP位点，我们首先在X染色体上鉴定出698个SNP位点，并筛选靠近X连锁基因Amelogenin（AMELX）及锌指蛋白（Zinc-finger protein, ZFX）的位点，这两个基因此前已被用于美洲豹及其他猫科动物的性别鉴定（Pilgrim等，2005）。随后我们分析每个个体的这些SNP基因型，并通过主成分分析可视化雌雄个体的位点分布情况。最终从中筛选出6个SNP位点，其基因型模式清晰：所有雌性个体均为纯合子，而雄性个体则为杂合子或携带另一种等位基因的纯合子。这些SNP位点分别位于X连锁基因ZFX、FRMPD4、TRAPPC2、TXLNG及USP9X上。