Low genetic diversity may be associated with an increase in species' extinction risk (Spielman et al. 2004, Frankham 2005). Still, global conservation assessments do not consider relevant genetic-based estimates for evaluating species threat status. Rather, they rely primarily on changes in population abundance and range size, with the inherent assumption that intra-specific genetic variability is tightly correlated with population size and range area (Frankham 1996). If this assumption was universally true, species considered to be at high risk, because of small range sizes and/or low abundances, should have lower levels of genetic diversity than low-risk species and vice-versa. However, contradictory evidence, for birds and mammals (Reed 2010), suggests that omitting genetic diversity from threat classification criteria could potentially lead to under- or over-estimating the actual extinction risk of species.

Here, we investigate whether bird species considered at risk of extinction, by widely used threat assessment criteria (IUCN 2021), have less intra-specific nucleotide diversity than non-threatened bird species (Supporting information). To accomplish this aim, we established differences in intra-specific nucleotide diversity for threatened (Vulnerable – VU, Endangered – EN and Critically Endangered – CR) vs non-threatened bird species (Least Concern – LC and Near Threatened – NT) by compiling 28 403 publicly available avian mitochondrial DNA (mtDNA) sequences from GenBank. We calculated cytochrome-b (cyt-b) nucleotide diversity for 1036 species (approximately 10% of all bird species), with an average number of sequences per species being 27 ± 44 (Supporting information). The average sequence length (base-pairs) across species was 887 ± 201. Using phylANOVAs, to control for phylogenetic signal (Freckleton et al. 2002), corrected for varying sample sizes between groups, we show that threatened species have significantly lower cyt-b nucleotide diversity (p < 0.05, in 953 out of the 1000 phylANOVA repetitions; mean p = 0.010 ± 0.025) than non-threatened species (Fig. 1a; Supporting information), with medium to large effect size in 97.2% of repetitions (ω² > 0.06). The mean effect size was 0.16 ± 0.05 (Supporting information).

Our results reveal that current threat assessment criteria indirectly prioritize species with low levels of cyt-b nucleotide diversity, which can be at greater risk of extinction by virtue of low genetic diversity (Frankham 2005) (Fig. 1c). For example, the African houbara (Chlamydotis undulata, VU) is among the birds with the lowest cyt-b nucleotide diversity in our data set (≤10th percentile: GD ≤ 0.0015; Fig. 1d), and its persistence is affected by inbreeding and/or genetic drift (Korrida et al. 2012). Moreover, the millerbird (Acrocephalus familiaris, CR) and the inaccessible finch (Nesospiza acunhae, VU) are both range-restricted small-island endemics with limited cyt-b nucleotide diversity (≤ 10th percentile; Supporting information), making them particularly vulnerable to rapid environmental changes from introduced predators and extreme climatic events (Vincenzi et al. 2017). Although mtDNA has been shown, under some circumstances, to be of limited use for inferring population size (Bazin et al. 2006, Nabholz et al. 2009), the low levels of nucleotide diversity in threatened species of birds suggest a correlation, direct or indirect, between cyt-b nucleotide diversity and small population or range size. For species that have not experienced large range contractions and population declines in recent times (non-threatened species), we found that cyt-b nucleotide diversity was generally high (≥90th percentile: GD ≥ 0.0302). Higher levels of genetic diversity might, through the process of local adaptation, aid species' resilience to rapid environmental changes (DeWoody et al. 2021) and reverse or slow species' decline (Fig. 1c). However, in some instances, non-threatened species can harbour low genetic diversity, most probably due to recent or past bottlenecks (Weber et al. 2000).

Four per cent of all non-threatened birds analysed had low levels of cyt-b nucleotide diversity (≤ 10th percentile; Fig. 1b). For example, the sooty tit (Aegithalos fuliginosus, LC; Fig. 1d) is the non-threatened species with the lowest cyt-b nucleotide diversity in our data set (Supporting information). Despite having a restricted range, the sooty tit is considered as ‘Least Concern', due to a population that is suspected to be stable (IUCN 2021). Low nucleotide diversity for the sooty tit signals that extinction risk for the species might be higher than its IUCN threat status indicates, encouraging further assessments of its conservation status using census and genomic techniques. Low genetic diversity in non-threatened species can result from recent or past dramatic demographic events, after which levels of intra-specific genetic diversity remain temporally low, while the overall population size increases (Weber et al. 2000). For these species, whole-genome studies will help reveal the role of genetic diversity in long-term species survival.

While our results could be contingent on the length of sequences, sample size, and geographic and taxonomic biases associated with genetic sequences in public repositories such as GenBank, we found no correlation between nucleotide diversity and average sequence length or number of sequences (Supporting information). Furthermore, we found a low phylogenetic signal (λ = 0.56, p < 0.001), and the phylANOVAs confirm the independence of the data in relation to the evolutionary history of the species (Supporting information). Indeed, there is a significant difference between the F-statistics calculated on the actual data and the F-statistics calculated with simulated data (null hypothesis; Supporting information). Lastly, our results do not reflect geographic biases in our dataset, which covers ~57% of all avian families and all zoogeographic realms (Supporting information). Despite existing challenges with using mitochondrial data and single genetic markers (Carling and Brumfield 2007), including the real possibility that genetic diversity calculated using mtDNA might not reflect genome-wide diversity or the diversity of specific functionally relevant parts of the genome, the relationship between conservation status and genetic diversity, explored in this paper, concords with long-standing expectations from the literature (DeWoody et al. 2021), including findings from meta-analyses across smaller subsets of taxa (Spielman et al. 2004, Willoughby et al. 2015) using nuclear DNA (allozymes, microsatellites, minisatellites), and other mtDNA genes (Petit-Marty et al. 2021).

Species-level conservation criteria capture low levels of intra-specific nucleotide variability in species of greatest concern. Nonetheless, low levels of nucleotide diversity are present in a small proportion of non-threatened birds, causing them, in theory, to be more vulnerable to rapidly changing environmental conditions than their conservation status, alone, indicates (Frankham 2005). As genomic techniques get cheaper, the inclusion of whole-genome data in relevant measures of genetic diversity is a likely near-term prospect for conservation. Future research should aim to integrate large-scale field-work campaigns with strategic sequencing of contemporary and historical specimens from biological collections, in order to unravel eco-evolutionary determinants of increased extinction risk.

Acknowledgements

– We thank Jonathan Kennedy for technical support and guidance. We thank Tim Melling for allowing us to use his photo of the sooty tit.

Funding

– This research was funded by Australian Research Council funding (grant no. FT140101192, DP180102392), awarded to DAF, the VILLUM FONDEN (grant no. 25925) awarded to CR, and the DFF project DEMOCHANGE (grant no. 8021-00282B) awarded to DNB.

Author contributions

Elisabetta Canteri: conceptualization (equal); data curation (equal); formal analysis (lead); writing - original draft (equal); writing – review and edit (equal). Damien A. Fordham: conceptualization (equal); funding acquisition (equal); supervision (equal); writing – review and edit (equal). Sen Li: investigation (lead); software (lead); writing – review and edit (equal). Peter A. Hosner: data curation (equal); writing – review and edit (equal). Carsten Rahbek: funding acquisition (equal); writing – review and edit (equal). David Nogues-Bravo: funding acquisition (equal); conceptualization (equal); supervision (lead); writing – original draft (equal); writing – review and edit (equal).

中文翻译：

---

IUCN 红色名录保护鸟类遗传多样性

低遗传多样性可能与物种灭绝风险的增加有关（Spielman et al. 2004 , Frankham 2005）。尽管如此，全球保护评估并未考虑相关的基于遗传的估计来评估物种威胁状况。相反，它们主要依赖于种群丰度和范围大小的变化，固有假设是种内遗传变异与种群大小和范围面积密切相关（Frankham 1996）。如果这一假设普遍成立，那么由于分布范围小和/或丰度低，被认为处于高风险的物种的遗传多样性水平应该低于低风险物种，反之亦然。然而，对于鸟类和哺乳动物的相互矛盾的证据（Reed 2010)，表明从威胁分类标准中忽略遗传多样性可能会导致低估或高估物种的实际灭绝风险。

在这里，我们通过广泛使用的威胁评估标准（IUCN 2021)，比未受威胁的鸟类具有更少的种内核苷酸多样性（支持信息）。为了实现这一目标，我们通过编译 28来自 GenBank 的 403 个公开可用的鸟类线粒体 DNA (mtDNA) 序列。我们计算了 1036 个物种（约占所有鸟类物种的 10%）的细胞色素 b (cyt-b) 核苷酸多样性，每个物种的平均序列数为 27 ± 44（支持信息）。跨物种的平均序列长度（碱基对）为 887 ± 201。使用 phylANOVAs，控制系统发育信号（Freckleton 等人，2002 年）)，针对组间不同的样本量进行校正，我们表明受威胁物种的 cyt-b 核苷酸多样性显着降低（p < 0.05，在 1000 次 phylANOVA 重复中的 953 次；平均 p = 0.010 ± 0.025）比未受威胁物种（图 1a；支持信息），在 97.2% 的重复中具有中到大的效应量（ω ² > 0.06）。平均效应大小为 0.16 ± 0.05（支持信息）。

我们的结果表明，当前的威胁评估标准间接优先考虑了 cyt-b 核苷酸多样性水平低的物种，由于遗传多样性低，这些物种可能面临更大的灭绝风险（Frankham 2005）（图 1c）。例如，非洲 houbara ( Chlamydotis undulata , VU) 是我们数据集中 cyt-b 核苷酸多样性最低的鸟类之一（≤ 10%：GD ≤ 0.0015；图 1d），其持久性受近交和/ 或遗传漂变（Korrida 等人，2012 年）。此外，水鹳 ( Acrocephalusfamiliais , CR) 和难以接近的雀科 ( Nesospiza acunhae, VU) 都是范围受限的小岛地方病，其 cyt-b 核苷酸多样性有限（≤ 10%；支持信息），使它们特别容易受到引入的捕食者和极端气候事件引起的快速环境变化的影响（Vincenzi 等人，2017 年） . 尽管在某些情况下，mtDNA 已被证明在推断种群规模方面的用途有限（Bazin 等人，2006 年，Nabholz 等人，2009 年）)，受威胁鸟类的核苷酸多样性水平低表明 cyt-b 核苷酸多样性与小种群或范围大小之间存在直接或间接的相关性。对于近期未经历大范围收缩和种群下降的物种（非受威胁物种），我们发现 cyt-b 核苷酸多样性普遍较高（≥90th 百分位：GD ≥ 0.0302）。通过局部适应过程，更高水平的遗传多样性可能有助于物种对快速环境变化的适应能力（DeWoody 等人，2021 年）并逆转或减缓物种的衰退（图 1c）。然而，在某些情况下，未受威胁的物种可能具有较低的遗传多样性，这很可能是由于最近或过去的瓶颈（Weber 等人，2000 年）。

分析的所有未受威胁鸟类中有 4% 的 cyt-b 核苷酸多样性水平较低（≤ 10%；图 1b）。例如，煤灰山雀（Aegithalos fuliginosus，LC；图 1d）是我们数据集中 cyt-b 核苷酸多样性最低的未受威胁物种（支持信息）。尽管范围有限，但由于怀疑种群数量稳定（IUCN 2021）。黑山雀的低核苷酸多样性表明该物种的灭绝风险可能高于其 IUCN 威胁状态所表明的，鼓励使用普查和基因组技术对其保护状况进行进一步评估。未受威胁物种的低遗传多样性可能是由于最近或过去的戏剧性人口统计事件造成的，此后种内遗传多样性的水平暂时保持低水平，而总体种群规模却在增加（Weber 等人，2000 年）。对于这些物种，全基因组研究将有助于揭示遗传多样性在物种长期生存中的作用。

虽然我们的结果可能取决于序列长度、样本大小以及与 GenBank 等公共存储库中的基因序列相关的地理和分类偏差，但我们发现核苷酸多样性与平均序列长度或序列数量之间没有相关性（支持信息） . 此外，我们发现了低系统发育信号（λ = 0.56，p < 0.001），phylANOVAs 证实了数据与物种进化历史相关的独立性（支持信息）。事实上，根据实际数据计算的 F 统计量与使用模拟数据计算的 F 统计量之间存在显着差异（零假设；支持信息）。最后，我们的结果不反映我们数据集中的地理偏差，涵盖了约 57% 的所有鸟类家族和所有动物地理领域（支持信息）。尽管在使用线粒体数据和单一遗传标记方面存在挑战（Carling 和 Brumfield2007 年），包括使用 mtDNA 计算的遗传多样性可能无法反映全基因组多样性或基因组特定功能相关部分的多样性的真实可能性，保护状态与遗传多样性之间的关系，在本文中探讨，与长期一致来自文献（DeWoody 等人，2021 年）的长期预期，包括使用核 DNA（同工酶、微卫星、小卫星）和其他分类群（Spielman 等人，2004 年，Willoughby 等人，2015 年）的荟萃分析结果mtDNA 基因（Petit-Marty 等人，2021 年）。

物种级别的保护标准捕获了最受关注的物种中低水平的特定内核苷酸变异性。尽管如此，在一小部分未受威胁的鸟类中存在低水平的核苷酸多样性，这导致它们理论上更容易受到快速变化的环境条件的影响，而不是它们的保护状态，单独表明（Frankham 2005）。随着基因组技术变得更便宜，将全基因组数据包含在遗传多样性的相关测量中可能是保护的近期前景。未来的研究应旨在将大规模实地工作活动与生物收藏中当代和历史标本的战略排序相结合，以解开导致灭绝风险增加的生态进化决定因素。

致谢

– 我们感谢 Jonathan Kennedy 的技术支持和指导。我们感谢蒂姆·梅林 (Tim Melling) 允许我们使用他的黑山雀照片。

资金

– 本研究由澳大利亚研究委员会资助（授予号 FT140101192、DP180102392）资助，授予 DAF、授予 VILLUM FONDEN（授予号 25925）和授予 DFF 项目 DEMOCHANGE（授予号 8021-00282B）到 DNB。

作者贡献

Elisabetta Canteri：概念化（平等）；数据管理（相等）；正式分析（铅）；写作 - 原稿（相等）；写作 - 审查和编辑（平等）。Damien A. Fordham：概念化（平等）；资金收购（平等）；监督（平等）；写作 - 审查和编辑（平等）。森立：调查（牵头）；软件（铅）；写作 - 审查和编辑（平等）。Peter A. Hosner：数据管理（平等）；写作 - 审查和编辑（平等）。Carsten Rahbek：资金收购（平等）；写作 - 审查和编辑（平等）。David Nogues-Bravo：资金收购（平等）；概念化（平等）；监督（牵头）；写作——原稿（相等）；写作 - 审查和编辑（平等）。