In 2020, the fire season affecting the western United States reached unprecedented levels. The 116 fires active in September consumed nearly 20,822 km² (https://inciweb.nwcg.gov/accessible-view/ Accessed 2020-09-29) with 80% of this footprint (16,567 km²) from 68 fires occurring within California, Oregon, and Washington. Although the 2020 fire season was the most extreme on record, it exemplified patterns of increased wildfire size, number, timing, return frequency, and extent, which are linked to climate-driven changes in precipitation and temperature affecting fire ignition and severity (Westerling 2016, Goss et al. 2020, Weber and Yadav 2020). In addition, wildfire smoke and particulate pollution have expanded greatly in recent decades throughout western North America, posing a threat to both human and ecological health (Burke et al. 2021). Wildfires have increasingly coincided with the start of fall migration (Westerling 2016, Goss et al. 2020) and may present a growing risk to migrating birds in the Pacific Flyway. Migrating birds across several western states were observed dead and dying in 2020. Within the Central Flyway, starvation of insectivorous birds that were recovered in Arizona, Colorado, and New Mexico was linked to a record cold-weather storm in the Rocky Mountains (Fox 2020). But causes of the nearly simultaneous bird mortalities of larger granivorous species (Fig. 1) further west in the Pacific Flyway, where fires were occurring, remain unclear.

Birds are especially vulnerable to disruption during migration as this is one of the most energetically demanding periods of the life cycle (McWilliams et al. 2004). Migration both follows and precedes periods of increased foraging activity and unanticipated energy deficits may have short-term carry-over effects on subsequent demographic processes (Newton 2006). Physiological and atmospheric constraints limit where, how high, and how long birds can fly. However, behavioral and physiological adaptations that make migration more efficient, such as using navigational cues, altering flight elevation to take advantage of winds, and efficient blood oxygenation are widespread across avian taxa (McWilliams et al. 2004). Some bird species fly in formation. which provides physical efficiencies (Mirzaeinia and Heppner 2020), but also maintains social cohesion within family groups, which is particularly strong in geese (Kölzsch et al. 2020). Many migrating species exhibit strong site fidelity to a limited number of stopover locations and travel routes (Weber and Houston 1999) potentially resulting in adverse demographic outcomes if traditional sites are not available (Klaassen et al. 2006). Such patterns of low “migratory diversity” are associated with greater probabilities of population decline while species with greater diversity may respond more robustly to climate change (Gilroy et al. 2016) or stochastic events. Temporary faculative migration from breeding ranges in advance of severe storms has been implied from light level loggers on songbirds that are otherwise obligate migrants (Streby et al. 2015), but a similar response was not evident for resident scavenging birds following volcanic eruption (Alarcón et al. 2016). It is less understood how actively migrating species respond to similar large-scale disruptions, particularly in the context of route finding and energy expenditure to traditionally used stopover regions.

Ecological investigations on the impact of wildfire on migration and movement have generally focused either on the energetic needs and foraging strategies of individuals or the impact of habitat or patch loss in the landscape. Wildfire impacts on actively migrating animals, particularly effects that extend beyond direct habitat loss within the fire perimeter, are not well documented due to the difficulty in obtaining migration information that spatially and temporally coincides with large fires. Beginning in 2018, we have marked tule greater white-fronted geese (Anser albifrons elgasi; tule goose) with collar-mounted GPS-enabled cellular transmitters to study movements and behaviors (Overton and Casazza 2021). With only 14,700 individuals (Yparraguirre et al. 2020), the tule goose is a Species of Special Concern in California and exhibits near absolute fidelity to a primary stopover site at the Summer Lake Wildlife Area in central Oregon, USA (96%; Yparraguirre et al. 2020). Typical fall migration, as observed from five radio-marked individuals in 2019, starts as a nearly direct route from the Cook Inlet in Alaska to Summer Lake that takes just over 4 d. The first half of fall migration between Alaska and Washington consists of flights over, and occasional short rests upon, the Pacific Ocean. Beginning September 12, 2020, all four tule geese were still transmitting, including a single individual that provided a migration track in 2019, and encountered dense smoke either while flying overland into southern Washington or while flying off the coast of Vancouver Island, British Columbia, Canada and Washington, USA.

The three-dimensional smoke forecast from NOAA's High-Resolution Rapid Refresh coupled with the Smoke (HRRR-Smoke) model (Ahmadov, 2017) were compared with bird movements to investigate any potential impacts of smoke upon migration patterns. Elevational gradients of smoke concentrations were linearly interpolated for every GPS location from each goose to allow interpretation of individual movement behavior relative to vertical distribution of smoke density. Geese responded to dense smoke concentrations averaging 161 µg m⁻³ (Fig. 2) by either stopping migration or altering direction and/or altitude of flights, which resulted in increased total flight time and distance. Three birds migrating over water stopped and rafted for 52–72 h before the smoke cleared and then moved inland. Landward migration of geese through smoke or directly over fires resulted in disorganized paths including both tangential and recursive flights, acute increases in altitude to 4,000 m to over-fly the smoke plume, and stopovers in non-traditional habitats occurring far from traditional migratory pathways (Fig. 3). Social cohesion of migrating flocks may also have been affected. Pairs of marked individuals migrated together twice, but groups separated each time that flocks were stopped in the smoke. One individual (Bird #192587) transited the fires but continued following prevailing winds within the smoke plume away from Summer Lake, arriving in Idaho where tule goose occurrence has never been confirmed (Fig. 3). Ultimately, all individuals arrived at Summer Lake after delays caused by smoke which more than doubled average migration duration (4.18 d in 2019 vs. 9.11 d in 2020; +118%) and extended the average flightpath during migration by 757 km (+27%).

Total energetic deficit resulting from delayed arrival at Summer Lake and extra distance flown was estimated using values from existing literature (see Appendix S1: Section S3). Total increased energetic expenditure averaged 950 kcals (3,977 kJ; observed range 404–1,118 kcals) which is equivalent to nearly 4.4 d of thermally neutral metabolism or 2.27 d encompassing normal activities during the winter. The duration of additional foraging needed to compensate for this caloric deficit is 27.5–42 h. Pre-breeding Canada geese (Branta canadensis) increase time spent feeding five-fold to nearly 31% of the day implying that even when feeding at behavioral extremes these energetic deficits would take 4–6 days before recovery. Energy deficits such as we describe, especially when occurring in the context of incomplete knowledge of available food resource locations, can lead to increased mortality or reproductive rates insufficient to maintain goose populations (Klaasen et al. 2005).

We observed impacts to individual migratory behavior at smoke concentrations averaging 161 µg m⁻³. At ground surface elevations, smoke concentrations exceeded this threshold across a geographic extent 27 times greater than the area burned by wildfires. However, at observed migration altitudes (<4,000 m), this smoke concentration covered an area 44 times larger than the wildfires themselves, encompassing 64% of four western states (California, Nevada, Oregon, and Washington) and effectively transecting the Pacific Flyway (Fig. 3).

Peak migration through the Pacific Flyway states of Washington, Oregon, and California, occurs when wildfires have become more intense, larger, and more frequent due to climatic variation and anthropogenic influences such as fire suppression and invasive plants (Goss et al. 2020, Weber and Yadav 2020). Wildfires in 2020 produced substantial barriers to bird navigation and movement across wide geographic extents, affecting route selection and energy expenditure sufficient to produce long-lasting effects. Limitations in high quality stopover sites due to continuing reductions in wetland extent within the Northern Great Basin (Donnelly et al. 2019) may exacerbate the impacts to migrating birds of larger and later occurring wildfires in western North America. Other species of geese, ducks, shorebirds, and passerines rely on similar traditional pathways and navigational cues and face comparable physiological and energetic constraints. The results we report for tule greater white-fronted geese may reflect significant challenges faced by a broader constituency of migrating birds, particularly those with low migratory diversity (Gilroy et al. 2016) and portend a future of increased challenges (Pickrell and Pennisi 2020) for migratory bird populations.

2020 年，影响美国西部的火灾季节达到了前所未有的水平。9 月发生的 116 起火灾消耗了近 20,822 公里² (https://inciweb.nwcg.gov/accessible-view/ 访问时间 2020-09-29)，其中 80% 的足迹 (16,567 公里² ) 来自加州发生的 68 起火灾、俄勒冈州和华盛顿州。尽管 2020 年的火灾季节是有记录以来最极端的，但它体现了野火规模、数量、时间、回归频率和范围增加的模式，这与气候驱动的降水和温度变化影响着火和严重程度有关（Westerling 2016 , Goss et al. 2020 , Weber and Yadav 2020）。此外，近几十年来，野火烟雾和颗粒物污染在整个北美西部急剧扩大，对人类和生态健康构成威胁（Burke et al. 2021）。野火越来越多地与秋季迁徙的开始同时发生（Westerling 2016，Goss et al. 2020），并可能对太平洋迁徙路线上的候鸟带来越来越大的风险。2020 年，观察到跨越几个西部州的候鸟死亡和死亡。在中央飞行路线内，在亚利桑那州、科罗拉多州和新墨西哥州恢复的食虫鸟类的饥饿与落基山脉创纪录的寒冷天气风暴有关（福克斯2020）。但是，在发生火灾的太​​平洋航道以西，较大的食肉物种（图 1）几乎同时发生鸟类死亡的原因仍不清楚。

鸟类在迁徙过程中特别容易受到破坏，因为这是生命周期中最需要精力的时期之一（McWilliams 等人，2004 年）。在觅食活动增加和未预料到的能量短缺时期之后和之前的迁移可能对随后的人口过程产生短期的遗留影响（Newton 2006）。生理和大气限制限制了鸟类可以飞到哪里、飞多高和飞多长时间。然而，使迁徙更有效的行为和生理适应，例如使用导航线索、改变飞行高度以利用风，以及有效的血液氧合在鸟类分类群中很普遍（McWilliams et al. 2004）。一些鸟类编队飞行。这提供了物理效率（Mirzaeinia 和 Heppner 2020），但也保持了家庭群体内的社会凝聚力，这在鹅中尤其强大（Kölzsch 等人2020）。许多迁徙物种对有限数量的中途停留地点和旅行路线表现出很强的站点保真度（Weber 和休斯顿，1999 年），如果传统站点不可用，可能会导致不利的人口统计结果（Klaassen 等人，2006 年）。这种低“迁徙多样性”模式与更大的人口下降概率有关，而具有更大多样性的物种可能对气候变化的反应更强劲（Gilroy 等人，2016 年）) 或随机事件。在强风暴之前从繁殖范围临时临时迁移已经从鸣禽的光级记录器中暗示，否则它们是专性迁徙的（Streby 等人，2015 年），但对于火山喷发后的常驻食腐鸟类来说，类似的反应并不明显（Alarcón 等人） al. 2016 年）。人们不太了解积极迁徙的物种如何应对类似的大规模中断，特别是在寻找路线和传统使用中途停留地区的能源消耗的背景下。

关于野火对迁徙和迁徙影响的生态调查通常集中在个体的能量需求和觅食策略或栖息地或景观中斑块丧失的影响。野火对积极迁徙的动物的影响，特别是超出火灾范围内直接栖息地丧失的影响，由于难以获得在空间和时间上与大火相吻合的迁徙信息，因此没有得到很好的记录。从 2018 年开始，我们已经标记了 tule 大白额鹅 ( Anser albifrons elgasi ; tule goose)，其项圈安装了支持 GPS 的蜂窝发射器，以研究运动和行为 (Overton 和 Casazza 2021 )。只有 14,700 人（Yparraguirre et al. 2020)，图勒鹅是加利福尼亚州的一种特别关注物种，对美国俄勒冈州中部夏湖野生动物区的主要中途停留地点表现出近乎绝对的忠诚度（96%；Yparraguirre 等人，2020）。根据 2019 年五个无线电标记个体观察到的典型秋季迁徙，从阿拉斯加库克湾到夏湖的几乎直接路线开始，仅需 4 天多一点。阿拉斯加和华盛顿之间秋季迁徙的前半部分包括飞越太平洋，偶尔短暂停留太平洋。从 2020 年 9 月 12 日开始，所有四只 tule 鹅仍在传播，其中包括 2019 年提供迁徙轨迹的单个个体，并且在陆路飞入华盛顿南部或飞离不列颠哥伦比亚省温哥华岛海岸时遇到浓烟，加拿大和美国华盛顿。

将 NOAA 高分辨率快速刷新与烟雾 (HRRR-Smoke) 模型 (Ahmadov, 2017 ) 相结合的三维烟雾预测与鸟类运动进行比较，以调查烟雾对迁徙模式的任何潜在影响。对每只鹅的每个 GPS 位置的烟雾浓度高程梯度进行线性插值，以解释相对于烟雾密度垂直分布的个体运动行为。^{鹅对平均 161 µg m -3}的浓烟浓度有反应（图2）通过停止迁移或改变飞行方向和/或高度，这导致总飞行时间和距离增加。三只在水上迁徙的鸟停下来漂流52-72小时，然后烟雾散去，然后向内陆移动。鹅通过烟雾或直接越过火灾向陆地迁移导致路径混乱，包括切向和递归飞行，高度急剧增加至 4,000 m 以飞越烟羽，以及在远离传统迁徙路径的非传统栖息地中途停留。图 3)。迁徙群体的社会凝聚力也可能受到影响。成对的标记个体一起迁移了两次，但每次群在烟雾中停下时，群体就会分开。一个人 (Bird #192587) 穿过了大火，但在远离夏日湖的烟羽中继续顺着盛行风到达爱达荷州，那里从未确认过 tule goose 的出现（图 3）。最终，所有个体在烟雾造成的延误后抵达夏湖，平均迁徙持续时间增加了一倍以上（2019 年 4.18 天，2020 年 9.11 天；+118%），迁徙期间的平均飞行路径延长了 757 公里（+27%） ）。

使用现有文献中的值估计了因延迟抵达夏湖和额外飞行距离而导致的总能量不足（见附录 S1：S3 节）。增加的总能量消耗平均为 950 kcal（3,977 kJ；观察范围为 404-1,118 kcal），相当于近 4.4 天的热中性代谢或 2.27 天（包括冬季正常活动）。补偿这种热量不足所需的额外觅食持续时间为 27.5-42 小时。预繁殖加拿大鹅 ( Branta canadensis )) 将每天进食的时间增加五倍，达到近 31%，这意味着即使在行为极端的情况下进食，这些能量不足也需要 4-6 天才能恢复。如我们所描述的能量不足，尤其是在对可用食物资源位置了解不完整的情况下，会导致死亡率增加或繁殖率不足以维持鹅的数量（Klaasen et al. 2005）。

^{我们观察到在平均 161 µg m -3}的烟雾浓度下对个体迁徙行为的影响。在地表海拔高度，烟雾浓度在地理范围内超过了这个阈值，是野火烧毁面积的 27 倍。然而，在观察到的迁移高度（<4,000 m），这种烟雾浓度覆盖的区域是野火本身的 44 倍，涵盖了西部四个州（加利福尼亚州、内华达州、俄勒冈州和华盛顿州）的 64%，并有效地横切了太平洋飞行路线（图 3)。

当由于气候变化和人为影响（例如灭火和入侵植物）而导致野火变得更加强烈、更大和更加频繁时，就会出现通过太平洋飞行路线州华盛顿、俄勒冈州和加利福尼亚州的迁移高峰（Goss 等人，2020 年，Weber和亚达夫2020 年）。2020 年的野火在广泛的地理范围内对鸟类导航和移动产生了巨大的障碍，影响了足以产生长期影响的路线选择和能量消耗。由于北部大盆地内湿地范围的持续减少，高质量的中途停留地点受到限制（Donnelly 等人，2019) 可能会加剧北美西部更大和后来发生的野火对候鸟的影响。其他种类的鹅、鸭、滨鸟和雀形目类动物依赖于类似的传统路径和导航线索，并面临类似的生理和能量限制。我们报告的 tule 大白额雁的结果可能反映了更广泛的候鸟群体面临的重大挑战，尤其是迁徙多样性低的鸟类（Gilroy 等人，2016 年），并预示着未来将面临更多挑战（Pickrell 和 Pennisi 2020 年）对于候鸟种群。