科学 · 林地蝙蝠

9800万次蝙蝠叫声或可揭示英格兰哪些林地需要加强保护

对英格兰310处林地的调查,正以前所未有的规模揭示蝙蝠活动情况。其保护价值取决于经过验证的物种记录,而不是把叫声当作种群数量。

播出 文字 阅读需 5 分钟19 个来源HeyDay News Desk · About usWatch report

通过 RSS 关注

原语言视频

文字实录 · 正在加载

译自英文原文。

新闻追踪第 1 天

1 篇报道8 家媒体19 个来源1 个视频

对于决定如何管理林地的人来说,知道某种易危蝙蝠会利用这片林地,可能会改变需要保护的对象。此前调查未发现的动物,仍可能依赖这种栖息地。这项在英格兰林地开展的大规模聆听调查,实际意义正在于此:揭示重要性被忽视的地点,而不只是收集更多野生动物声音。

国家林地蝙蝠调查已从310处林地调查点收集到超过98 million次蝙蝠叫声录音,研究人员借助人工智能,在217 TB的录音中检测和识别物种。Forest Research和伦敦大学学院于2026年10月30日公布了这些数据。调查规模令人瞩目;真正的保护考验在于,研究人员能否据此可靠地得出结论。10 9

关键事实

  1. 超过98 million记录的是蝙蝠叫声,而非个体动物数量 109
  2. 310接受调查的英格兰林地地点 109
  3. 217 TB收集的录音数据量 10

BBC新闻报道称,在此前没有记录过宽耳蝠的地区发现了这种蝙蝠,尤其是在英格兰西南部。蝙蝠保护信托将宽耳蝠(Barbastella barbastellus)在大不列颠和英格兰均列为易危物种。确认发现这种稀有物种,可能会促使自然保护人士进一步调查一片林地。1 18

但现有证据并未指出这些报道所述发现发生在哪些具体林地。项目的主要信息确认了调查的总体规模,却未独立证实新记录的宽耳蝠集中出现在西南部这一情况。因此,这一地理发现应继续注明来源为BBC,而不应被表述为已发布的、详细展示该物种分布地点的地图。1 10

新记录不等于种群恢复

发现一只蝙蝠与证明其生存状况发生变化,两者之间的区别很重要。首次记录只能说明研究人员在那里检测到了这种动物,并不能说明它是最近才来到那里。更有效的聆听方式,可能会发现此前调查漏掉的野生动物。

单独一次新发现也不能证明这种蝙蝠正在繁殖、分布范围正在扩大,或种群数量正在增长。这些是不同的问题,需要更多证据。调查报告的新发现为了解栖息地利用情况提供了有希望的线索,但并不能证明宽耳蝠正在恢复。

对叫声总数的解读也需要谨慎。一只蝙蝠可能反复发声,因此超过98 million次叫声不能换算成超过98 million只蝙蝠。声学记录有助于研究人员调查活动情况和识别物种;但记录本身并不是种群普查。

项目的精确总数也与BBC最初所述的“超过90 million次叫声,来自300多片林地”相符。这些数字是下限,并非相互矛盾的统计结果。尽管项目名称和部分公告采用了全国或英国范围的表述,但这310处地点只是英格兰林地的抽样,并不等于覆盖整个英国,也不代表英格兰的每一片林地。1 10 14

BBC报道称,设备连续录音超过三年。这段时间的确切起止日期尚未确定。另据该项目的数据目录,合作自2021年起开展;不应把这个日期误认为已确认的三年录音期起始时间。12 14

从自动标签到证据

项目由Forest Research和蝙蝠保护信托牵头,资金来自环境、食品与乡村事务部的自然资本与生态系统评估计划。伦敦大学学院(UCL)也参与其中。面对如此庞大的录音量,完全靠人工分析会极其费力。自动化处理让这些档案更便于使用。3 9 12

录音通过蝙蝠保护信托的声音分类系统进行处理。人工智能检测蝙蝠声音,并借助声谱图中的模式辅助按物种分类;声谱图是展示声音频率和时间变化的图像。现有文档无法确认具体使用的底层分类器或软件版本,因此没有依据在此为结果标注特定模型名称或准确率百分比。8 1

关键一步是核验:把机器生成的标签与经过确认的生物学记录区分开来。对于稀有物种,这不是无关紧要的技术细节。某片林地是否被认为有易危蝙蝠栖息,可能取决于研究人员能否确认录到的叫声确实来自这种蝙蝠。

该项目的2024年示例数据集明确包含未经验证的自动蝙蝠回声定位叫声分类结果。这一描述适用于该示例;它并不能证明整个调查的每项发现都未经验证。另有一个数据集被编目为“已验证的蝙蝠活动”。同样,这些经核验记录的存在,也不意味着全部98 million次叫声都经过逐一验证。整个调查中经验证的比例仍不清楚。3 14

正因如此,具体记录的质量比庞大的总量更重要。一条记录若明确记载了某物种在特定林地被发现,便可用于栖息地评估。数量再庞大的不确定标签,也不会仅仅因为数量多就发挥同样的作用。

Forest Research表示,英国所有蝙蝠都依赖林地觅食、栖息或往返于不同地点。因此,这项调查的潜在价值不仅在于发现稀有物种:更清楚地了解哪些蝙蝠利用不同林地,有助于识别重要栖息地,并为未来的管理提供指引。16

目前尚未证明这项调查带来了具体的管理方式变化、新的保护地认定,或蝙蝠种群经测量确认的改善。它目前取得的成果,是大幅提升了聆听蝙蝠声音的能力。要将这一成果转化为保护行动,就需要明确联系起经过验证的物种记录、这些动物利用的林地,以及对这些林地作出的管理决策。

因此,接下来值得关注的是具体信息:哪些林地发现了新记录的宽耳蝠,这些发现如何得到核验,以及管理者会采取哪些不同做法。与又一个令人瞩目的总数相比,这些答案对蝙蝠更重要。

播出原貌 14 行
  1. Recorders captured more than 98 million bat calls in England's woodlands. AI is helping researchers identify species and potentially overlooked habitat for vulnerable bats. How much can those recordings tell us about which woodlands need protection?
  2. The National Woodland Bat Survey collected 217 terabytes of recordings from 310 woodland sites. Those totals sharpen the earlier description of more than 90 million calls from more than 300 woodlands. The earlier figures were lower bounds, so the more precise numbers fit within them.
  3. An individual bat can call repeatedly. Researchers can use those sounds to investigate activity and identify species, but turning call totals into animal numbers would give a misleading picture of how many bats live in these woods.
  4. The survey covers English woodlands, with devices recording over three years. Barbastelles have reportedly been detected in previously unrecorded areas, especially south-west England. The 310 sites represent a sample of English woodland, rather than equivalent coverage across the whole United Kingdom.
  5. Forest Research and the Bat Conservation Trust lead the project. Defra funds it through its Natural Capital and Ecosystem Assessment programme. The partnership has operated since 2021. That partnership date is separate from the three-year recording period, whose exact beginning and end haven't been specified.
  6. The barbastelle is classified as Vulnerable in both Great Britain and England. Finding this rare bat in a woodland where its presence had gone unrecorded could change how that habitat's importance is understood, giving conservationists a reason to examine it more closely.
  7. A first record tells us that a bat was detected there. Better listening can reveal animals that earlier surveys missed. Researchers would need other evidence to show that barbastelles are moving into new territory, breeding there, or increasing in number. The new detections alone can't demonstrate population recovery.
  8. AI detects bat sounds and helps classify calls by species using patterns in spectrograms. These are pictures of sound across frequency and time. The Bat Conservation Trust's Sound Classification System processes the recordings, while verification separates an automated label from a checked biological record.
  9. University College London, known as UCL, is involved in the project. At this scale, analysing the audio by hand would be extremely laborious. Automated processing makes the collection more practical to use, but the conservation value depends on how reliably researchers identify the species behind the calls.
  10. The project's 2024 example dataset explicitly contains unvalidated automated call classifications. For a rare species, checking a label has practical consequences. A woodland's apparent importance could depend on whether researchers can confirm that the call really belongs to that bat.
  11. A separate dataset contains Verified Bat Activity records. The example's unvalidated status applies to that example, while the verified dataset provides checked activity evidence. Its existence doesn't mean all 98 million calls were individually checked. The survey-wide share verified isn't clear.
  12. All UK bats depend on woodland for feeding, roosting or travelling between places. That makes the survey useful beyond the rare-species discoveries. A clearer picture of which species use different woods can help researchers identify important habitat and guide future woodland management.
  13. The recordings can reveal where bats are active and help identify habitat that deserves closer attention. Turning that information into conservation decisions requires clearly documented species records. The survey's practical promise rests on connecting verified detections with the woods those animals use.
  14. The next details needed are which woodlands contain the newly recorded barbastelles, how those detections were checked, and what management decisions will follow. No specific management change or measured improvement in bat populations has yet been demonstrated by this survey.

来源

19 个来源8 家媒体1 个视频1 篇通稿

来源地图

追溯本篇报道的依据。图谱展示其关联;下方链接可打开本文引用的来源。

来源图谱将在您阅读时加载。

探索完整的新闻编辑室图谱 ↗

译文更新于

修订记录

  1. r1首次发布。