Boreal ecosystems are experiencing extensive changes because of anthropogenic stressors such as climate change. Information on density of species at multiple sites is vital to understand and manage the impact of these changing conditions on boreal forest communities. Yet, for most boreal forest species, including the vast majority of mammals, obtaining reliable estimates of density is exceedingly difficult. Recently developed methods for the estimation of densities of unmarked animals from camera-trapping data could help to overcome this hurdle, but have not yet been empirically validated in many ecosystems. Here, we assess the ability of camera traps to estimate density of snowshoe hare (Lepus americanus) using three different models: the random encounter model (REM), the random encounter and staying time (REST) model, and the time-to-event (TTE) model. We additionally evaluate the relationship between hare density and two simple indices based on camera detection rate and pellet counts. Across 13 sites in North Central Washington, United States, we compared live-trapping spatially explicit capture–recapture (SECR) estimates of density to the three camera-based density models and the two indices. We found that the camera-based models, in particular the REM and REST models, performed well in estimating densities consistent with the live-trapping data, with an average difference in density from SECR-based estimates of only 0.12 and 0.13 hares/ha, respectively. Both indices also had strong predictive relationships with hare density. Our results show that, owing to their noninvasive nature and relative ease of application, camera-based methods could be used to obtain hare density estimates at much larger spatiotemporal scales than have been applied to date. Given the keystone role of hare in boreal ecosystems, and emerging evidence of hare range retraction, the ability to estimate densities across many sites is a key tool for hare conservation and management. Moreover, our results are highly encouraging for the application of camera-based methods to obtain density estimates on a wide variety of boreal forest species, though additional validation will be necessary.

中文翻译：

---

使用相机陷阱估计雪鞋野兔 (Lepus americanus) 的密度：一种重要的北方森林食草动物

由于气候变化等人为压力因素，北方生态系统正在经历广泛的变化。多个地点的物种密度信息对于理解和管理这些不断变化的条件对北方森林群落的影响至关重要。然而，对于大多数北方森林物种，包括绝大多数哺乳动物来说，获得可靠的密度估计是极其困难的。最近开发的从相机捕获数据估计未标记动物密度的方法可以帮助克服这一障碍，但尚未在许多生态系统中得到经验验证。在这里，我们使用三种不同的模型评估相机陷阱估计雪鞋野兔（Lepus americanus）密度的能力：随机遭遇模型（REM）、随机遭遇和停留时间（REST）模型，和事件发生时间 (TTE) 模型。我们还根据相机检测率和颗粒计数评估了野兔密度与两个简单指标之间的关系。在美国华盛顿中北部的 13 个地点，我们将实时捕获的空间显式捕获-再捕获 (SECR) 密度估计值与三个基于相机的密度模型和两个指数进行了比较。我们发现基于摄像机的模型，尤其是 REM 和 REST 模型，在估计与实时捕获数据一致的密度方面表现良好，与基于 SECR 的估计值的平均密度差异仅为 0.12 和 0.13 野兔/公顷，分别。这两个指数也与野兔密度有很强的预测关系。我们的研究结果表明，由于它们的非侵入性和相对易于应用，基于相机的方法可用于在比迄今为止应用的更大的时空尺度上获得野兔密度估计。鉴于野兔在北方生态系统中的关键作用，以及野兔范围收缩的新证据，估计许多地点的密度的能力是野兔保护和管理的关键工具。此外，我们的结果对于应用基于相机的方法来获得各种北方森林物种的密度估计是非常令人鼓舞的，尽管还需要额外的验证。