Search for: All records

Every night during spring and autumn, the mass movement of migratory birds redistributes bird abundances found on the ground during the day. However, the connection between the magnitude of nocturnal migration and the resulting change in diurnal abundance remains poorly quantified. If departures and landings at the same location are balanced throughout the night, we expect high bird turnover but little change in diurnal abundance (stream‐like migration). Alternatively, migrants may move simultaneously in spatial pulses, with well‐separated areas of departure and landing that cause significant changes in the abundance of birds on the ground during the day (wave‐like migration). Here, we apply a flow model to data from weather surveillance radars (WSR) to quantify the daily fluxes of nocturnally migrating birds landing and departing from the ground, characterizing the movement and stopover of birds in a comprehensive synoptic scale framework. We corroborate our results with independent observations of the diurnal abundances of birds on the ground from eBird. Furthermore, we estimate the abundance turnover, defined as the proportion of birds replaced overnight. We find that seasonal bird migration chiefly resembles a stream where bird populations on the ground are continuously replaced by new individuals. Large areas show similar magnitudes of take‐off and landing, coupled with relatively small distances flown by birds each night, resulting in little change in bird densities on the ground. We further show that WSR‐inferred landing and take‐off fluxes predict changes in eBird‐derived abundance turnover rate and turnover in species composition. We find that the daily turnover rate of birds is 13% on average but can reach up to 50% on peak migration nights. Our results highlight that WSR networks can provide real‐time information on rapidly changing bird distributions on the ground. The flow model applied to WSR data can be a valuable tool for real‐time conservation and public engagement focused on migratory birds' daytime stopovers. 

More than two billion birds migrate through the Gulf of Mexico each spring en route to breeding grounds in the USA and Canada. This region has a long history of complex natural and anthropogenic environments as the northern Gulf coast provides the first possible stopover habitats for migrants making nonstop trans‐Gulf crossings during spring migration. However, intense anthropogenic activity in the region, which is expanding rapidly at present, makes migrants vulnerable to a multitude of obstacles and increasingly fragments and alters these habitats. Understanding the timing of migrants' overwater arrivals has biological value for expanding our understanding of migration ecology relative to decision‐making for nonstop flights, and is imperative for advancing conservation of this critical region through the identification of key times in which to direct conservation actions (e.g. temporary halting of wind turbines, reduction of light pollution). We explored 10 years of weather surveillance radar data from five sites along the northern Gulf of Mexico coast to quantify the daily timing and intensity of arriving trans‐Gulf migrants. On a daily scale, we found that migrant intensity peaked an average of nine hours after local sunrise, occurring earliest at easternmost sites. On a seasonal level, the greatest number of arrivals occurred between late April and early May, with peak intensity occurring latest at westernmost sites. Overall intensity of migration across all 10 years of data was greatest at the westernmost sites and decreased moving farther to the east. These findings emphasize the differential spatial and temporal patterns of use of the Gulf of Mexico region by migrating birds, information that is essential for improving our understanding of the ecology of trans‐Gulf migration and for supporting data‐driven approaches to conservation actions for the migratory birds passing through this critical region. 

Abstract Weather radar networks have great potential for continuous and long-term monitoring of aerial biodiversity of birds, bats, and insects. Biological data from weather radars can support ecological research, inform conservation policy development and implementation, and increase the public’s interest in natural phenomena such as migration. Weather radars are already used to study animal migration, quantify changes in populations, and reduce aerial conflicts between birds and aircraft. Yet efforts to establish a framework for the broad utilization of operational weather radar for biodiversity monitoring are at risk without suitable data policies and infrastructure in place. In Europe, communities of meteorologists and ecologists have made joint efforts toward sharing and standardizing continent-wide weather radar data. These efforts are now at risk as new meteorological data exchange policies render data useless for biodiversity monitoring. In several other parts of the world, weather radar data are not even available for ecological research. We urge policy makers, funding agencies, and meteorological organizations across the world to recognize the full potential of weather radar data. We propose several actions that would ensure the continued capability of weather radar networks worldwide to act as powerful tools for biodiversity monitoring and research. 

Class imbalance in the training data hinders the generalization ability of machine listening systems. In the context of bioacoustics, this issue may be circumvented by aggregating species labels into super-groups of higher taxonomic rank: genus, family, order, and so forth. However, different applications of machine listening to wildlife monitoring may require different levels of granularity. This paper introduces TaxoNet, a deep neural network for structured classification of signals from living organisms. TaxoNet is trained as a multitask and multilabel model, following a new architectural principle in end-to-end learning named "hierarchical composition": shallow layers extract a shared representation to predict a root taxon, while deeper layers specialize recursively to lower-rank taxa. In this way, TaxoNet is capable of handling taxonomic uncertainty, out-of-vocabulary labels, and open-set deployment settings. An experimental benchmark on two new bioacoustic datasets (ANAFCC and BirdVox-14SD) leads to state-of-the-art results in bird species classification. Furthermore, on a task of coarse-grained classification, TaxoNet also outperforms a flat single-task model trained on aggregate labels. 

To explain the consonance of octaves, music psychologists represent pitch as a helix where azimuth and axial coordinate correspond to pitch class and pitch height respectively. This article addresses the problem of discovering this helical structure from unlabeled audio data. We measure Pearson correlations in the constant-Q transform (CQT) domain to build a K-nearest neighbor graph between frequency subbands. Then, we run the Isomap manifold learning algorithm to represent this graph in a three-dimensional space in which straight lines approximate graph geodesics. Experiments on isolated musical notes demonstrate that the resulting manifold resembles a helix which makes a full turn at every octave. A circular shape is also found in English speech, but not in urban noise. We discuss the impact of various design choices on the visualization: instrumentarium, loudness mapping function, and number of neighbors K.