Avian influenza virus is a causal agent of avian influenza, which is belonging to Orthomyxoviridae, and possess the two types of glycoproteins on the surface of a virion; hemagglutinin (HA) and neuraminidase (NA). Based on the reactivity to antiserum, the HA and NA of Type A influenza viruses are classified into H1 to H16, and N1 to N9, respectively. Since 1997, high pathogenicity avian influenza (HPAI), especially caused by the infection of H5HA, had been reported in Asia. Especially, since 2010, H5N8 HPAI viruses (HPAIVs) have been prevalent not only in Asia but also in other regions including Europe and Africa. As of 2020, the H5N8s were further classified into clade 2.3.4.4 according to the amino acid sequence, and H5 HPAIV subclades of a to h in the clade 2.3.4.4 are confirmed. Graduate School of Veterinary Medicine was designated as OIE reference laboratory for highly and low pathogenic avian influenza since 2005 and had conducted several actions including global surveillance of avian influenza. We had already stocked the isolates and developed the genetic database of them (https://virusdb.czc.hokudai.ac.jp).

In October 2020, we isolated the H5N8 HPAIV from a fecal sample collected at the Komuke Lake in Hokkaido under the global surveillance, which was designated as A/northern pintail/Hokkaido/M13/2020 (H5N8) (NP/Hok/20). It was revealed that NP/Hok/20 was classified into the H5 clade 2.3.4.4b in the HA of the phylogenetic tree and genetically very close to the H5N8 HPAIV isolated in Europe in the winter of 2019-2020. Furthermore, comprehensive phylogenetical analysis using the H5N8 HPAIV isolated in the Far East in the winter of 2020-2021 revealed that those isolates were classified into three categories including 1) the European isolates in the winter of 2019-2020, 2) the European isolates in the winter of 2020-2021, and 3) isolates in China and Mongolia. Since the genetic reassortment between the European H5N8 isolates and avian influenza viruses in the Asia was confirmed in the phylogenetical analysis, further spread with wider genetic variation might occur.

HPAI spread situations in the Europe and Japan in the end of 2020 were similar to ones in the end of 2016; many of the contagious viruses were brought by bird migration from the northern territories. Given the situation that very close viruses were isolated at both edges of the Eurasian continent at the same season, it is likely that the contagious viruses had been already perpetuated in the northern territory where the migratory birds nest in the summer season. Periodic updates of intensive survey at the global level as well as the reinforcement of the biosecurity measures in the poultry farm are essential to prepare for future HPAI outbreak.

Avian influenza (AI) surveillance in wild birds had been conducted all over the world, as migratory water birds are worrying potential carriers of highly pathogenic avian influenza (HPAI) viruses. Risk assessment of AI in wild birds had been conducted intensively in European and American countries, and target species and high priority areas for AI surveillance have been specified. Nowadays, passive surveillance of dead and debilitated birds are major surveillance methods in most of these countries. National HPAI surveillance in wild birds was started in Japan in 2008. Passive surveillance of reported dead birds and active surveillance of waterbird feces were conducted in each prefecture as part of the national surveillance. The livestock hygiene service centers in most of the prefectures conducted influenza rapid diagnostic tests. AI surveillance in wild birds is essential not only to perceive infection status in wild birds, but to provide important information for rare bird conservation, both in the wild and in captivity. In addition, the early detection of HPAI infection in wild birds plays an important role in the alert of poultry and captive birds. Surveillance system in wild birds should be prepared to maintain at any situation, such as in the middle of severe outbreaks of livestock diseases.

本研究では，大規模酪農場で飼養されている乳牛を対象として，1日当たり分娩数と気象条件および月相の関連性を分析すること，夜間分娩に関連する要因を明らかにすること，そして潮位と分娩の関連性を明らかにすることを目的とした。九州に所在するホルスタイン種約2,500頭を飼養する大規模酪農場1農場を対象とし，2013年1月1日から2016年12月31日までの1,461日間に分娩した8,485分娩記録を収集した。1日当たり分娩数と気象条件および月相の関連性に関して，調査に用いた最高気温などの気象条件および月相は1日当たり分娩数と有意な関連性がみられなかった。夜間（19時から7時前まで）における分娩確率に関して，夜間分娩の割合は46.3%であり，昼間の時間帯よりも少なかった。夜間分娩確率は初産牛において2産以上の牛よりも高く，夜間分娩のオッズ比は1.11倍高かった（95%信頼区間：1.01-1.22）。また，1-3月分娩と比較して，7-9月分娩では夜間分娩のオッズ比が1.16倍高かった（95%信頼区間：1.03-1.31）。潮位と分娩数の関連性に関して，初産牛では，潮位が高くなるにつれて1時間当たり分娩数が有意に減った（回帰係数±標準誤差：0.00086±0.000362；P＝0.01）。一方，2産以上の牛では潮位と分娩数の関連性はみられなかった。まとめとして，乳牛の分娩と気象条件および月相の関連性はみられなかったが，夜間分娩の確率は初産牛および7-9月分娩で高くなった。また，初産牛では潮位と分娩の関連性がみとめられた。

Evaluation of regression models is essential for identifying avoidable errors in the data manipulation process and model building, and assessing the robustness of model results. The model evaluation has two steps; assessment of the overall goodness of fit and identifying influential observations. This article covers a fundamental concept of the model evaluation and provides a few strategies to deal with poorly fitting models. A tutorial material using R is developed where readers can replicate line-by-line a model building and evaluation process.