1. The timing of birth has a predominant influence on both the reproductive success of the mother and the life-history trajectory of her offspring. Because early growth and survival are key drivers of population dynamics, there is an urgent need to understand how global change is affecting reproductive phenology and performance. However, identifying when and where birth occurs is often difficult in the wild due to the cryptic behaviour of females around parturition, although this information may also help managers to protect reproductive females and newborn against human disturbance. While several approaches to identify parturition based on movement metrics derived from GPS monitoring have previously been proposed, their performance has not been evaluated over a range of species with contrasted movement characteristics.
2. Here, we present a novel approach to detect parturition by combining data on animal movements, activity rate and habitat use. Using machine learning approaches, we evaluated the relative and combined performance of each category of metrics in predicting parturition for three large herbivores with contrasted life histories: a hider-type species, the roe deer Capreolus capreolus and two follower-type species, the Mediterranean mouflon Ovis gmelini musimon × Ovis sp. and the Alpine ibex Capra ibex.
3. We first showed that detection of parturition was much improved when birth-related modifications in the habitat use and activity rate of the mother were considered, rather than relying on movement metrics only. We then demonstrated that our approach was highly successful (76%–100% of events correctly identified) in detecting parturition in both follower and hider species. Furthermore, our approach generated estimates for peak birth date and the proportion of parturient females that were comparable with those based on direct observations at the population scale. Finally, our approach outperformed the most commonly employed methods in the literature which generally failed to identify non-reproductive females for the three studied species, and provided birth timing estimates that only poorly match the true parturition date.
4. We suggest that by combining sources of information, we have developed a standardised methodological approach for inferring parturition in the wild, not only for large herbivores but also for any species where parturition induces marked behavioural changes in the mother.

中文翻译：

---

推断分娩的标准化生物记录方法：在大型食草动物中的应用

1. 出生时间对母亲的生殖成功和后代的生活史轨迹都有主要影响。由于早期生长和存活是种群动态的关键驱动因素，因此迫切需要了解全球变化如何影响生殖物候和性能。然而，由于雌性在分娩前后的神秘行为，在野外通常很难确定分娩发生的时间和地点，尽管这些信息也可能有助于管理人员保护生殖雌性和新生儿免受人为干扰。虽然之前已经提出了几种基于 GPS 监测得出的运动指标来识别分娩的方法，但尚未在一系列具有对比运动特征的物种上评估它们的性能。
2. 在这里，我们提出了一种通过结合动物运动、活动率和栖息地使用数据来检测分娩的新方法。使用机器学习方法，我们评估了每类指标在预测三种具有对比生活史的大型食草动物的分娩方面的相对和综合性能：一种躲藏型物种，狍狍狍和两种追随型物种，地中海欧洲欧洲盘羊Ovis gmelini musimon  ×  Ovis sp. 和高山山羊山羊山羊山羊。
3. 我们首先表明，当考虑与出生相关的栖息地使用和母亲活动率的改变时，分娩的检测得到了很大改善，而不是仅依赖于运动指标。然后我们证明了我们的方法在检测跟随者和隐藏者物种的分娩方面非常成功（76%–100% 的事件正确识别）。此外，我们的方法生成了峰值出生日期和产妇比例的估计值，这些估计值与基于人口规模的直接观察结果相当。最后，我们的方法优于文献中最常用的方法，这些方法通常无法识别三个研究物种的非生殖雌性，并且提供的出生时间估计与真正的分娩日期不符。
4. 我们建议，通过结合信息来源，我们开发了一种标准化的方法论方法来推断野外分娩，不仅适用于大型食草动物，而且适用于分娩导致母体发生显着行为变化的任何物种。