UTILIZAÇÃO DE APRENDIZADO PROFUNDO PARA CLASSIFICAÇÃO ENTRE DIFERENTES ESPÉCIES DE ACCIPITRÍDEOS

Introdução: Accipitrídeos são aves voadoras membros da família Accipitridae, cujos representantes mais simbólicos são as águias, alguns gaviões e os abutres do velho mundo. Para fins de classificação, neste trabalho escolhemos as seguintes águias: Harpia (Harpia harpyja), águia-careca (Haliaeetus leucocephalus), águia-dourada(Aquila chrysaetos) e águia-das-filipinas (Pithecophaga jefferyi). Dentre essas, a águia-das filipinas e a harpia encontram-se ameaçadas devido a destruição do seu habitat e caça ilegal. Assim, ferramentas que possibilitam classificação rápida desses animais seriam importantes para auxiliar pesquisadores e leigos na diferenciação entre os accipitrídeos viventes. Objetivo: Construir um classificador de imagens que seja capaz de diferenciar entre imagens dos vários tipos de accipitrídeos com pequeno gasto computacional. Materiais e métodos: Para realizar o trabalho foi escolhida a 47 atualização do dataset “315-birds”, disponível da plataforma Kaggle, e com base no dataset foram separadas as espécies da família accipitridae disponíveis, e foi-se utilizado a proporção de divisão já pré-determinada pelo no treino/teste/validação. Para a inteligência artificial utilizou-se do pacote Keras integrado ao api Tensor Flow,e foi definida a rede neural Mobile-Net. Para gerar o modelo, treinou-se por 80 épocas em 10 seeds diferentes (327, 261, 777, 47, 954, 728, 901, 553, 419, 934). Por fim, ao fim de cada seed foi coletado as médias das métricas estabelecidas (precisão, recall e f1-score), e se calculou média geral e desvio padrão. Resultados: A avaliação do modelo retornou os seguintes resultados: recall 92%±4,8% , precisão 92,3%±4,1% e f1-score 91,8%±5,4%. O accipeterídeo que apresentou maior relativa dificuldade de ser classificado foi a Harpia (Harpia harpyja), que apresentou métricas de precisão média de 84,6%±8,45%, que mesmo com a dificuldade relatada pode ser considerado um valor aceitável. Conclusão: Conclui-se que o modelo é capaz de atingir os objetivos esperados, e assim capaz de classificar de forma satisfatória os diferentes tipos de accipitrídeos com grande confiabilidade. Além disso, o trabalho demonstra que a computação é capaz de distinguir espécies anatomicamente similares, e que pode ser expandido para outros trabalhos envolvendo outras aves ou animais diversos.

Rotenone Use and Subsequent Prey Loss Lowers Osprey Fledging Rates Via Brood Reduction

ABSTRACT Fisheries managers used the fish toxicant rotenone to eradicate an undesirable brown bullhead (Ameiurus nebulosus) population and all other fish species at Hyatt Reservoir, Oregon, on 12 October 1989. This 4-yr study (1988–1990, 1992) compared effects of that rotenone project on Ospreys (Pandion haliaetus) nesting at Hyatt Reservoir and nearby Howard Prairie Reservoir (untreated reference)—the latter a reservoir where both brown bullheads and hatchery-released rainbow trout (Oncorhynchus mykiss) prospered. Because Hyatt Reservoir was treated after Osprey fall migration in 1989, the first 2 yr (1988 and 1989) yielded pretreatment information: number of Osprey pairs was unchanged and reproductive rates were similar and consistent at the two reservoirs. Yearling fish (200–250 mm) were restocked at Hyatt Reservoir in the spring of 1990 and Ospreys returned each year following rotenone treatment, with no decline in the number of occupied or active nests. The negative effect of the rotenone treatment on Ospreys was short-term, resulting in reduced reproductive rates (young/occupied nest, young/active nest, and young/successful nest) during the first nesting season posttreatment, although hatching rates were not affected. Osprey dive success and prey delivery rates declined sharply in 1990, leading to competition for food among siblings and brood reduction. Osprey reproductive rates and prey delivery rates at Hyatt Reservoir in both 1990 and 1992 remained below the extremely high pretreatment rates, but within the range required for population stability. Serious adverse effects of the fish loss on Osprey reproduction were minimized by: (1) the delay of the rotenone application until after breeding season, (2) the restocking of the treated reservoir in the following spring with some larger (yearling) fish (though the timing was late), (3) the maintenance of a supplemental feeding program for a nesting pair of Bald Eagles (Haliaeetus leucocephalus), which minimized kleptoparasitism on Ospreys, and perhaps most important (4) the presence of nearby water bodies, where Osprey obtained some fish in the 1990 and 1992 breeding seasons.

Nest Distribution of Four Priority Raptor Species In Colorado

ABSTRACT Raptors face threats such as habitat modification, climate change, and environmental pollutants in many parts of the western USA, where rapid human population growth exacerbates such pressures. However, information about distribution of raptor nests at broad spatial scales that could inform conservation efforts is lacking. To provide a contemporary estimate of nest distribution of four raptor species of special conservation concern (Bald Eagle [Haliaeetus leucocephalus], Ferruginous Hawk [Buteo regalis], Golden Eagle [Aquila chrysaetos], and Prairie Falcon [Falco mexicanus]) throughout Colorado, we used a statewide database of raptor nesting locations to inform species distribution models for monitoring and management efforts. We used generalized linear models to identify the relationship between nest locations and explanatory covariates relating to land cover, temperature, topography, and prey distribution. We investigated the effect of different methods for selecting the sample of locations available to raptors, comparing four selection frames: sampling from the observed locations of the target-group (i.e., other raptor nests), sampling from within a spatial buffer around observed locations, sampling from outside of the same buffer, or complete random sampling of the background locations without respect to observations. Out-of-sample validation techniques indicated strong predictive accuracy of our models. Each raptor species was best represented by a different one of the four approaches to sample available locations, refuting our expectation that models accounting for bias would perform better than those that did not. Our findings were consistent with generally understood habitat associations of these species. These models can be used to identify hot spots with high relative probability of use by breeding raptors and to inform future monitoring practices that use a standardized, stratified sampling design.

Ideal free eagles: Bald Eagle distribution in relation to salmon availability on four salmon spawning rivers

The movement of individuals according to the availability of resources has a fundamental effect on animal distributions. In the Pacific Northwest, Bald Eagles (Haliaeetus leucocephalus (Linneaus, 1766)) rely heavily on scavenging opportunities during the non-breeding period, and their distribution and movements are thought to be strongly influenced by the availability of post-spawning Pacific salmon (genus Oncorhynchus Suckley, 1861) carcasses. We surveyed the abundance of eagles and salmon on four adjacent rivers on Vancouver Island, British Columbia, Canada, during the 2017 fall spawning season. Salmon began to arrive in late September, peaked in abundance in mid-November, and were absent after early December. The seasonal progression of Bald Eagle abundance matched that of salmon carcass availability. The slope of proportional eagle – salmon relationship was significantly positive, though lower than the 1:1 match predicted by Ideal Free Distribution theory. The numerical response of Bald Eagles to salmon abundance was elevated on one of the rivers, potentially due to physical features such as sandbars and mudflats that increased the availability of carcasses and provided access points for eagles.

Anticoagulant rodenticide exposure and toxicosis in bald eagles (Haliaeetus leucocephalus) and golden eagles (Aquila chrysaetos) in the United States

Raptors, including eagles, are geographically widespread and sit atop the food chain, thereby serving an important role in maintaining ecosystem balance. After facing population declines associated with exposure to organochlorine insecticides such as dichlorodiphenyltrichloroethane (DDT), bald eagles (Haliaeetus leucocephalus) have recovered from the brink of extinction. However, both bald and golden eagles (Aquila chrysaetos) are exposed to a variety of other toxic compounds in the environment that could have population impacts. Few studies have focused on anticoagulant rodenticide (AR) exposure in eagles. Therefore, the purpose of this study was to determine the types of ARs that eagles are exposed to in the USA and better define the extent of toxicosis (i.e., fatal illness due to compound exposure). Diagnostic case records from bald and golden eagles submitted to the Southeastern Cooperative Wildlife Disease Study (University of Georgia) 2014 through 2018 were reviewed. Overall, 303 eagles were examined, and the livers from 116 bald eagles and 17 golden eagles were tested for ARs. The percentage of AR exposure (i.e., detectable levels but not associated with mortality) in eagles was high; ARs were detected in 109 (82%) eagles, including 96 (83%) bald eagles and 13 (77%) golden eagles. Anticoagulant rodenticide toxicosis was determined to be the cause of mortality in 12 (4%) of the 303 eagles examined, including 11 bald eagles and 1 golden eagle. Six different AR compounds were detected in these eagles, with brodifacoum and bromadiolone most frequently detected (81% and 25% of eagles tested, respectively). These results suggest that some ARs, most notably brodifacoum, are widespread in the environment and are commonly consumed by eagles. This highlights the need for research to understand the pathways of AR exposure in eagles, which may help inform policy and regulatory actions to mitigate AR exposure risk.

Perkembangan Terkini Investigasi Gen Leptin pada Ayam: Ulasan

Perkembangan investigasi gen pengkode leptin (LEP) pada beberapa spesies golongan aves telah dilakukan secara mendalam selama kurun waktu 20 tahun. Ekspresi gen LEP pada genom beberapa spesies aves telah teridentifikasi diantaranya Falco peregrinus dan Falco cherrug, Pseudopodoces humilis, Taeniopygia guttata, Columba livia, Haliaeetus leucocephalus, Picoides pubescens, dan Melopsittacus undulatus. Namun identifikasi genetika molekuler gen LEP pada ayam (G. gallus gallus) masih diperdebatkan. Perkembangan industri peternakan ayam lokal asli Indonesia tergolong pesat, namun masih memiliki beberapa keterbatasan salah satunya produktivitas ternak ayam lokal asli Indonesia. Pemenuhan kebutuhan pangan nasional salah satunya berasal dari sektor peternakan unggas khususnya ayam lokal asli Indonesia dengan kontribusi sebesar 23% per tahun. Peningkatan produktivitas peternakan ayam lokal asli Indonesia dapat dicapai dengan persilangan selektif. Persilangan selektif berupaya dalam meningkatkan mutu genetik ternak ayam lokal asli Indonesia dengan memberdayakan plasma nutfah ayam lokal asli Indonesia. Persilangan selektif memungkinkan dihasilkannya hibrida ayam pedaging dan petelur dengan karakteristik unggul sesuai dengan rancangan pola persilangan yang ditetapkan. Dalam proses persilangan selektif dibutuhkan penerapan genetika Mendelian dan molekuler sebagai panduan dan evaluasi arah seleksi guna meningkatkan efektivitas persilangan. Beberapa gen yang berpengaruh signifikan terhadap produktivitas ayam telah berhasil diidentifikasi, namun beberapa kandidat gen penting lainnya seperti gen LEP masih membutuhkan riset yang mendalam. Ulasan ini mengintegrasikan riset gen LEP pada ayam dari awal penemuan hingga terobosan terkini.