How changing landscapes reshape migration and avian influenza risk景观变化如何重塑迁徙与禽流感风险

First, the landscape changes the birds首先，景观变化改变了水鸟的分布

The starting point was a study led by Dr. Yanjie Xu (University of Helsinki) that did not examine avian influenza at all. Instead, we asked how habitat degradation affected the seasonal distributions of eight migratory waterfowl species in the East Asian-Australasian Flyway (EAAF). Using bird observations, species distributions and land-cover data, we mapped changes in wetland availability, fragmentation and isolation between 1992 and 2012 [1].这个故事的起点，可以回溯到一项由赫尔辛基大学徐燕洁博士领衔、但本身并不涉及禽流感的研究。我们关注的是，栖息地退化如何影响东亚-澳大利西亚迁飞区（EAAF）8 种迁徙水鸟的季节性分布。通过整合鸟类观测、物种分布和土地覆盖数据，我们描绘了 1992 至 2012 年间湿地可利用性、破碎化和隔离程度的变化 [1]。

The changes were strongly uneven across the flyway. Wetlands generally deteriorated more in the southern and eastern parts of the flyway, including coastal China and Japan, while habitat conditions improved in some inland and northern regions. Importantly, species were not affected equally. Waterfowl with longer, narrower migration corridors were more exposed to habitat degradation because they had fewer parallel alternatives when important sites were lost. Species using shorter and broader corridors had more flexibility to shift among alternative sites [1].结果显示，这些变化在整个迁飞区并不均匀。中国沿海和日本等迁飞区南部、东部地区的湿地总体退化更明显，而一些内陆和北部地区的栖息地条件则有所改善。更重要的是，不同物种受到的影响并不相同。迁徙通道更长、更狭窄的水鸟，在关键地点丧失后可选择的平行替代栖息地更少，因此更容易暴露于栖息地退化的影响。相比之下，迁徙通道较短、较宽的物种，在不同替代地点之间调整的空间更大 [1]。

That result is fundamentally about conservation and migration ecology, not disease. But it raises an obvious next question: if landscape change can redirect migratory birds and alter where they aggregate, how does this change influence the spatial and temporal dynamics of avian influenza transmission that vectored by these migratory birds?这项研究讨论的核心是保护生物学和迁徙生态，并不是疾病生态学的相关后果。但它自然引出了下一个问题，如果景观变化能够改变迁徙路线和鸟类聚集的位置，那么它又会怎样影响这些迁徙水鸟参与传播的禽流感病毒的时空传播格局？

Migration can change infection risk in more than one way迁徙对感染风险的影响，并不只有一个方向

Landscape change matters for disease partly because it reshapes the migration routes and stopover sites that determine where birds aggregate and which epidemiological environments they encounter. To understand this link, we first examined how different migration routes relate to HPAI exposure. In one empirical study, we compared the migration routes of Swan Geese and Bar-headed Geese, together with two Whooper Swan populations, against domestic duck densities and reported HPAI outbreaks. Swan Geese actually used areas with higher domestic duck densities and were closer to outbreak locations, yet no HPAI outbreaks were recorded in Swan Geese in the dataset, whereas Bar-headed Geese had multiple recorded outbreaks [2].景观变化之所以可能影响疾病，一个重要原因是它会重新塑造迁徙路线和停歇地，从而改变鸟类在哪里聚集、又会经过怎样的流行病学环境。为了理解这条联系，我们首先从经验数据出发，比较不同迁徙路线与 HPAI 暴露之间的关系。在一项研究中，我们比较了鸿雁和斑头雁的迁徙路线，并加入两个大天鹅种群作为对照，同时叠加家鸭密度和已报告的 HPAI 疫情。结果有些出乎意料：鸿雁活动区域的家鸭密度更高，而且迁徙过程中距离疫情地点更近，但数据集中没有记录到鸿雁 HPAI 疫情，相反，斑头雁则有多次疫情记录 [2]。

Exposure alone therefore could not explain the contrasting infection records. We proposed two possible explanations: frequent historical contact with domestic ducks might be associated with greater immune protection in Swan Geese, whereas the energetic and physiological costs of high-altitude migration might increase vulnerability in Bar-headed Geese. These were hypotheses rather than directly tested mechanisms, but the contrast showed that migration route, exposure and host response need to be considered together [2].因此，单纯的“暴露更多”并不能解释这些物种之间不同的感染记录。我们提出了两种可能的解释：鸿雁长期、频繁接触家鸭，可能与更强的免疫保护有关，而斑头雁进行高海拔迁徙所付出的能量和生理成本，则可能增加其感染脆弱性。这些都只是尚未被直接检验的机制假设，但这个对比说明，理解疾病风险时，需要把迁徙路线、暴露机会和宿主反应放在一起考虑 [2]。

Landscape change can also alter the structure of migration itself by removing, adding or redistributing stopover opportunities. To isolate what such changes in migration structure might mean for infection, we used a theoretical transmission model to test how migration network configuration and timing affect AIV dynamics. We compared networks containing serial or parallel stopover sites and varied how synchronized the birds were in their movements. Multiple serial stopovers and highly synchronized migration reduced cumulative infection in the model, partly because birds repeatedly moved into sites where environmental virus levels were still low. In contrast, parallel stopovers could accelerate infection accumulation, and less synchronized migration increased cumulative infection because later birds encountered environments already contaminated by earlier arrivals [3].景观变化还可能通过移除、增加或重新分布停歇地，直接改变迁徙网络本身。为了单独理解这种“迁徙结构变化”可能怎样影响感染，我们进一步建立了一个理论传播模型，检验迁徙网络结构和迁徙时间对 AIV 动态的影响。我们比较了依次使用多个停歇地的串联网络和分流使用多个停歇地的并行网络，同时改变鸟群迁徙的同步程度。模型中，更多串联停歇地和更高的迁徙同步性降低了累计感染，因为鸟群可以不断进入环境病毒水平仍较低的新地点。相反，并行停歇地可能加快感染累积；当迁徙同步性降低时，较晚到达的个体更容易遇到已被先到个体污染的环境，因此累计感染增加 [3]。

In other words, migration can facilitate pathogen dispersal, but it can also allow birds to move away from contaminated site-a process known as “migratory escape.” Whether migration amplifies or reduces infection depends on how the migration network is organized and how birds move through it [3].换句话说，迁徙既可以促进病原体的远距离扩散，也可以帮助宿主离开已经积累病毒的地点，这就是所谓的“迁徙逃逸”（migratory escape）。迁徙究竟会放大还是降低感染风险，取决于迁徙网络如何组织，以及鸟类如何在网络中移动 [3]。

What happens when habitats disappear?当栖息地不断消失，会发生什么？

We then connected these ideas directly to landscape change using an agent-based model. For Greater White-fronted Geese, we constructed six fall-migration networks representing progressively greater habitat loss and simulated both individual migration and AIV transmission. As sites were removed, the migration network became smaller, birds shifted their distribution, and larger congregations formed at the remaining sites [4].接下来，我们用基于个体的模型把这些机制直接连接到景观变化。以白额雁为例，我们构建了 6 个代表栖息地逐步丧失的秋季迁徙网络，并同时模拟个体迁徙和 AIV 传播。随着迁徙站点不断被移除，整个网络变得更小，鸟类的空间分布发生改变，剩余地点则聚集了更多个体 [4]。

Those movement changes altered the modeled epidemiology. Simulated outbreak risk shifted northward with the birds, and remaining sites experienced higher outbreak risk because more geese were concentrated there. Habitat loss also increased virus transmission through greater congregation and environmental contamination. Under the intact network, migration could reduce the main infection peak through migratory escape; extensive habitat loss shortened migration and weakened that protective effect [4].这些迁徙变化进一步改变了模型中的流行病学过程。模拟中的疫情风险随着鸟类分布一起向北移动，而剩余地点由于聚集了更多雁类，疫情风险随之升高。栖息地丧失还通过更大的鸟群聚集和环境中的病毒积累，促进了病毒传播。在完整迁徙网络中，“迁徙逃逸”可以降低主要感染高峰；但严重的栖息地丧失缩短了迁徙过程，也削弱了这种保护效应 [4]。

From virus transmission to viral diversification从病毒传播进一步走向病毒多样化

Our most recent model extended the same framework from transmission within wild birds to the wild bird-poultry interface. We combined satellite tracking, eBird observations, remote-sensing data and poultry distributions to reconstruct landscapes representing 2000 and 2015. The model allowed Greater White-fronted Geese to respond to changes in wetlands, surface water and rice paddies, and then simulated transmission of a wild-bird-origin AIV strain into poultry already carrying a poultry-origin strain [5].在我们最新的模型中，这一框架又从野鸟种群内部的传播，扩展到了野鸟-家禽界面。我们整合卫星追踪、eBird 观测、遥感数据和家禽分布，重建了代表 2000 年和 2015 年的两种景观情景。模型中的白额雁会根据湿地、地表水和稻田的变化调整迁徙和栖息地利用，随后，我们模拟野鸟来源的 AIV 毒株向家禽种群溢出，而这些家禽中同时存在一个家禽来源的 AIV 毒株 [5]。

This matters because coinfection creates an opportunity for viral reassortment, the exchange of genome segments between influenza viruses. In the model, landscape changes between 2000 and 2015 reshaped bird migration, increased wild-poultry spillover and produced substantially more reassortment in poultry. The study reported a substantial increase in reassortment rate, with simulated risk increasing across parts of southeastern China, the Yellow River Basin and northeastern China [5].这里关键的一步是“双重感染”。当同一个宿主同时感染两种流感病毒时，就可能发生病毒重配，也就是不同病毒之间交换基因组片段。在模型中，2000 至 2015 年间的景观变化重塑了水鸟迁徙，增加了野鸟向家禽的溢出传播，也使家禽中的病毒重配明显增加。模拟显示，中国东南部、黄河流域和东北地区的部分区域，其重配风险都有所上升 [5]。

That result should be interpreted carefully. We did not simulate the complete molecular evolution of HPAI, and reassortment was used as a proxy for ecological and transmission conditions that could favor viral diversification and the emergence of novel subtypes. The model therefore identifies a mechanism and potential risk landscape, not a prediction that a particular new HPAI strain will emerge at a particular place or time [5].但这一结果需要谨慎理解。我们并没有模拟 HPAI 完整的分子进化过程，而是把重配作为一个代理指标，用来表示更有利于病毒多样化和新亚型出现的生态与传播条件。因此，这个模型识别的是一种可能的作用机制及其潜在风险空间格局，而不是预测某一种新的 HPAI 毒株会在某个具体地点、某个具体时间出现 [5]。

A changing landscape is part of the epidemiological system景观变化，也是流行病学系统的一部分

Taken together, these studies trace a continuous pathway. Landscape change alters habitat availability. Habitat availability reshapes bird distribution and migration networks. Those movement changes alter aggregation, environmental exposure and contact with poultry. These changes can in turn affect virus transmission, spillover and opportunities for reassortment [1–5].把这些研究放在一起看，可以看到一条连续的线索。景观变化首先改变栖息地可利用性。栖息地变化进一步重塑鸟类分布和迁徙网络。迁徙变化又会改变鸟类聚集、环境暴露以及与家禽接触的机会。最终，这些变化可能进一步影响病毒传播、跨物种溢出和病毒重配的机会 [1-5]。

The implication is not that every habitat change increases disease risk. Habitat loss can concentrate birds, while habitat gains can attract birds into new areas, including landscapes shared with domestic poultry. What matters is where change occurs, how it alters connectivity, and which new contacts it creates. For surveillance, this means that disease hotspots should not be treated as fixed points on a map. Combining landscape monitoring, bird tracking and virus surveillance can help us follow a flyway as it changes, and identify where ecological change may create the next transmission opportunity.当然，这并不意味着任何一种栖息地变化都会提高疾病风险。栖息地丧失可能迫使鸟类集中到更少的地点，而新的或改善的栖息地也可能把鸟类吸引到新的区域，包括与家禽共享的景观。真正重要的是，变化发生在哪里？它怎样改变迁徙网络的连通性？又创造了哪些新的接触机会？因此，在疾病监测中，高风险地点不应被视为地图上固定不变的点。把景观监测、鸟类追踪和病毒监测结合起来，可以帮助我们跟踪一个不断变化的迁飞区，并识别哪些生态变化可能创造新的病毒传播机会。