ABSTRACT. Historic population trajectories for most North American bird species are largely unknown for years prior to circa 1970. Additionally, current estimates of population trajectories of boreal and Arctic breeding species are imprecise or biased because of lack of coverage by Breeding Bird Survey (BBS) routes in that region. Citizen science data, in particular eBird data, could fill these information gaps. Bayesian regression models of eBird data were used to estimate population trajectories of 22 boreal or Arctic breeding species of songbirds, 4 migratory songbird species that breed in eastern North America, and 2 species of raptors whose populations crashed due to the pesticide DDT. Models used range-wide data from the U.S. and Canada for spring migration/breeding, fall migration, and winter. To evaluate the model results, comparisons were made between eBird models from different seasons, between eBird indices and area defoliated by spruce budworm (Choristoneura fumiferana), and between eBird, BBS, and Christmas Bird Count (CBC) annual indices and trends. Population trajectories were positively correlated between seasons for most of the species analyzed based on correlations between annual indices, magnitude of trends, and residuals from trend models. Of the species analyzed, those most often associated with spruce budworm outbreaks had the strongest correlations between eBird annual indices and area defoliated by spruce budworm in the boreal forest. Annual indices from eBird models were positively correlated with BBS for most species, and trends calculated through the annual indices from eBird models were strongly correlated with those from the BBS for spring (r = 0.73, n = 25, P < 0.0001), fall (r = 0.64, n = 25, P = 0.0005), and winter (r = 0.81, n = 9, P = 0.0084), and winter eBird trends were correlated with those from the CBC (r = 0.64, n = 12, P = 0.0252). The results suggest eBird analyses could be an important complement to the BBS, CBC, and other surveys for assessing the status of bird species in North America, and that historic population trajectories could be estimated with additional historic eBird checklists.

中文翻译：

---

评估 eBird 数据对北美鸟类历史种群轨迹建模以及监测北方和北极繁殖物种种群的有效性

摘要。大多数北美鸟类的历史种群轨迹在 1970 年左右之前的数年中在很大程度上是未知的。此外，由于缺乏繁殖鸟类调查 (BBS) 路线的覆盖，目前对北方和北极繁殖物种种群轨迹的估计不准确或有偏差。那个地区。公民科学数据，尤其是 eBird 数据，可以填补这些信息空白。eBird 数据的贝叶斯回归模型用于估计 22 种北方或北极繁殖鸣禽、4 种在北美东部繁殖的迁徙鸣禽以及 2 种因杀虫剂 DDT 而种群数量减少的猛禽的种群轨迹。模型使用来自美国和加拿大的范围广泛的数据进行春季迁徙/繁殖、秋季迁徙和冬季迁徙。为了评估模型结果，对不同季节的 eBird 模型、eBird 指数和云杉幼虫 (Choristoneura fumiferana) 落叶面积以及 eBird、BBS 和圣诞鸟类计数 (CBC) 年度指数和趋势进行了比较。根据年度指数、趋势幅度和趋势模型残差之间的相关性，分析的大多数物种的种群轨迹在季节之间呈正相关。在分析的物种中，那些最常与云杉芽虫爆发相关的物种在 eBird 年度指数与北方森林中云杉芽虫落叶面积之间的相关性最强。对于大多数物种，eBird 模型的年度指数与 BBS 呈正相关，通过 eBird 模型的年度指数计算的趋势与来自 BBS 的春季 (r = 0.73, n = 25, P < 0.0001)、秋季 (r = 0.64, n = 25, P = 0.0005) 和冬季（r = 0.81，n = 9，P = 0.0084），冬季 eBird 趋势与来自 CBC 的趋势相关（r = 0.64，n = 12，P = 0.0252）。结果表明，eBird 分析可以作为 BBS、CBC 和其他评估北美鸟类状况的调查的重要补充，并且可以通过额外的历史 eBird 清单来估计历史种群轨迹。