RESUMO
We present a new method of detecting North Atlantic Right Whale (NARW) upcalls using a Multimodel Deep Learning (MMDL) algorithm. A MMDL detector is a classifier that embodies Convolutional Neural Networks (CNNs) and Stacked Auto Encoders (SAEs) and a fusion classifier to evaluate their output for a final decision. The MMDL detector aims for diversity in the choice of the classifier so that its architecture is learned to fit the data. Spectrograms and scalograms of signals from passive acoustic sensors are used to train the MMDL detector. Guided by previous applications, we trained CNNs with spectrograms and SAEs with scalograms. Outputs from individual models were evaluated by the fusion classifier. The results obtained from the MMDL algorithm were compared to those obtained from conventional machine learning algorithms trained with handcrafted features. It showed the superiority of the MMDL algorithm in terms of the upcall detection rate, non-upcall detection rate, and false alarm rate. The autonomy of the MMDL detector has immediate application to the effective monitoring and protection of one of the most endangered species in the world where accurate call detection of a low-density species is critical, especially in environments of high acoustic-masking.
Assuntos
Aprendizado Profundo , Baleias , Acústica , Algoritmos , Animais , Redes Neurais de ComputaçãoRESUMO
Key to predicting likely consequences of future climate change for Arctic marine mammals is developing a detailed understanding of how these species use their environment today and how they were affected by past climate-induced environmental change. Genetic analyses are uniquely placed to address these types of questions. Molecular genetic approaches are being used to determine distribution and migration patterns, dispersal and breeding behavior, population structure and abundance over time, and the effects of past and present climate change in Arctic marine mammals. A review of published studies revealed that population subdivision, dispersal, and gene flow in Arctic marine mammals was shaped primarily by evolutionary history, geography, sea ice, and philopatry to predictable, seasonally available resources. A meta-analysis of data from 38 study units across seven species found significant relationships between neutral genetic diversity and population size and climate region, revealing that small, isolated subarctic populations tend to harbor lower diversity than larger Arctic populations. A few small populations had substantially lower diversity than others. By contrast, other small populations retain substantial neutral diversity despite extensive population declines in the 19th and 20th centuries. The evolutionary and contemporary perspectives gained from these studies can be used to model the consequences of different climate projections for individual behavior and population structure and ultimately individual fitness and population viability. Future research should focus on: (1) the use of ancient-DNA techniques to directly reconstruct population histories through the analysis of historical and prehistorical material, (2) the use of genomic technologies to identify, map, and survey genes that directly influence fitness, (3) long-term studies to monitor populations and investigate evolution in contemporary time, (4) further Arctic-wide, multispecies analyses, preferably across different taxa and trophic levels, and (5) the use of genetic parameters in population and species risk analyses.
Assuntos
Clima , Ecologia , Mamíferos , Biologia Marinha , Animais , Mamíferos/genéticaRESUMO
The Major Histocompatibility Complex (MHC) is a critical element in mounting an effective immune response in vertebrates against invading pathogens. Studies of MHC in wildlife populations have typically focused on assessing diversity within the peptide binding regions (PBR) of the MHC class II (MHC II) family, especially the DQ receptor genes. Such metrics of diversity, however, are of limited use to health risk assessment since functional analyses (where changes in the PBR are correlated to recognition/pathologies of known pathogen proteins), are difficult to conduct in wildlife species. Here we describe a means to predict the binding preferences of MHC proteins: We have developed a model positional scanning library analysis (MPSLA) by harnessing the power of mixture based combinatorial libraries to probe the peptide landscapes of distinct MHC II DQ proteins. The algorithm provided by NNAlign was employed to predict the binding affinities of sets of peptides generated for DQ proteins. These binding affinities were then used to retroactively construct a model Positional Scanning Library screen. To test the utility of the approach, a model screen was compared to physical combinatorial screens for human MHC II DP. Model library screens were generated for DQ proteins derived from sequence data from bottlenose dolphins from the Indian River Lagoon (IRL) and the Atlantic coast of Florida, and compared to screens of DQ proteins from Genbank for dolphin and three other cetaceans. To explore the peptide binding landscape for DQ proteins from the IRL, combinations of the amino acids identified as active were compiled into peptide sequence lists that were used to mine databases for representation in known proteins. The frequency of which peptide sequences predicted to bind the MHC protein are found in proteins from pathogens associated with marine mammals was found to be significant (p values <0.0001). Through this analysis, genetic variation in MHC (classes I and II) can now be associated with the binding repertoires of the expressed MHC proteins and subsequently used to identify target pathogens. This approach may be eventually applied to evaluate individual population and species risk for outbreaks of emerging diseases.
Assuntos
Alelos , Golfinhos/genética , Biblioteca Gênica , Antígenos de Histocompatibilidade Classe II/genética , Modelos Genéticos , Regiões Promotoras Genéticas , Animais , Golfinhos/imunologia , Antígenos de Histocompatibilidade Classe II/imunologia , ProteômicaRESUMO
Climate change is having profound impacts on animal populations, and shifts in geographic range are predicted in response. Shifts that result in range overlap between previously allopatric congeneric species may have consequences for biodiversity through interspecific competition, hybridization, and genetic introgression. Harbor seals (Phoca vitulina) and spotted seals (Phoca largha) are parapatric sibling species and areas of co-occurrence at the edges of their range, such as Bristol Bay, Alaska, offer a unique opportunity to explore ecological separation and discuss potential consequences of increased range overlap resulting from retreating sea ice. Using telemetry and genetic data from 14 harbor seals and six spotted seals, we explored the ecological and genetic separation of the two species by comparing their utilization distributions, distance from haul-out, dive behavior (e.g., depth, duration, focus), and evidence of hybridization. Firstly, we show that harbor and spotted seals, which cannot be visually distinguished definitively in all cases, haul-out together side by side in Bristol Bay from late summer to early winter. Secondly, we observed subtle rather than pronounced differences in ranging patterns and dive behavior during this period. Thirdly, most spotted seals in this study remained close to shore in contrast to what is known of the species in more northern areas, and lastly, we did not find any evidence of hybridization. The lack of distinct ecological separation in this area of sympatry suggests that interspecific competition could play an important role in the persistence of these species, particularly if range overlap will increase as a result of climate-induced range shifts and loss of spotted seal pagophilic breeding habitat. Our results also highlight the added complexities in monitoring these species in areas of suspected overlap, as they cannot easily be distinguished without genetic analysis. Predicted climate-induced environmental change will likely influence the spatial and temporal extent of overlap in these two sibling species. Ultimately, this may alter the balance between current isolating mechanisms with consequences for species integrity and fitness.