BioCPPNet: automatic bioacoustic source separation with deep neural networks

We introduce the Bioacoustic Cocktail Party Problem Network (BioCPPNet), a lightweight, modular, and robust U-Net-based machine learning architecture optimized for bioacoustic source separation across diverse biological taxa. Employing learnable or handcrafted encoders, BioCPPNet operates directly on the raw acoustic mixture waveform containing overlapping vocalizations and separates the input waveform into estimates corresponding to the sources in the mixture. Predictions are compared to the reference ground truth waveforms by searching over the space of (output, target) source order permutations, and we train using an objective function motivated by perceptual audio quality. We apply BioCPPNet to several species with unique vocal behavior, including macaques, bottlenose dolphins, and Egyptian fruit bats, and we evaluate reconstruction quality of separated waveforms using the scale-invariant signal-to-distortion ratio (SI-SDR) and downstream identity classification accuracy. We consider mixtures with two or three concurrent conspecific vocalizers, and we examine separation performance in open and closed speaker scenarios. To our knowledge, this paper redefines the state-of-the-art in end-to-end single-channel bioacoustic source separation in a permutation-invariant regime across a heterogeneous set of non-human species. This study serves as a major step toward the deployment of bioacoustic source separation systems for processing substantial volumes of previously unusable data containing overlapping bioacoustic signals.

Comparative Genomics Analysis of Repetitive Elements in Ten Gymnosperm Species: “Dark Repeatome” and Its Abundance in Conifer and Gnetum Species

Repetitive elements (RE) and transposons (TE) can comprise up to 80% of some plant genomes and may be essential for regulating their evolution and adaptation. The “repeatome” information is often unavailable in assembled genomes because genomic areas of repeats are challenging to assemble and are often missing from final assembly. However, raw genomic sequencing data contain rich information about RE/TEs. Here, raw genomic NGS reads of 10 gymnosperm species were studied for the content and abundance patterns of their “repeatome”. We utilized a combination of alignment on databases of repetitive elements and de novo assembly of highly repetitive sequences from genomic sequencing reads to characterize and calculate the abundance of known and putative repetitive elements in the genomes of 10 conifer plants: Pinus taeda, Pinus sylvestris, Pinus sibirica, Picea glauca, Picea abies, Abies sibirica, Larix sibirica, Juniperus communis, Taxus baccata, and Gnetum gnemon. We found that genome abundances of known and newly discovered putative repeats are specific to phylogenetically close groups of species and match biological taxa. The grouping of species based on abundances of known repeats closely matches the grouping based on abundances of newly discovered putative repeats (kChains) and matches the known taxonomic relations.

Revisiting the influence of learning in predator functional response, how it can lead to shapes different from type III.

Predator/Parasitoid functional response is one of the main tools used to study predation behaviour, and in assessing the potential of biological control candidates. It is generally accepted that predator learning in prey searching and manipulation can produce the appearance of type III functional response. Holling proposed that in the presence of alternative prey, at some point the predator would shift the preferred prey, leading to the appearance of a sigmoid function that characterized that functional response. This is supported by the analogy between enzyme kinetics and functional response that Holling used as the basis for developing this theory. However, after several decades, sigmoidal functional responses appear in the absence of alternative prey in most of the biological taxa studied. Here, we propose modelling the effect of learning on the functional response by using the explicit incorporation of learning curves in the parameters of the Holling functional response, the attack rate (a), and the manipulation time (h). We then study how the variation in the parameters of the learning curves causes variations in the shape of the functional response curve. We found that the functional response product of learning can be either type I, II or III, depending on what parameters act on the organism, and how much it can learn throughout the length of the study. Therefore the presence of other types of curves should not be automatically associated with the absence of learning. These results are important from an ecological point of view because when type III functional response is associated with learning, it is generally accepted that it can operate as a stabilizing factor in population dynamics. Our results, to the contrary, suggest that depending on how it acts, it may even be destabilizing by generating the appearance of functional responses close to type I.

BioCPPNet: Automatic Bioacoustic Source Separation with Deep Neural Networks

We introduce the Bioacoustic Cocktail Party Problem Network (BioCPPNet), a lightweight, modular, and robust UNet-based machine learning architecture optimized for bioacoustic source separation across diverse biological taxa. Employing learnable or handcrafted encoders, BioCPPNet operates directly on the raw acoustic mixture waveform containing overlapping vocalizations and separates the input waveform into estimates corresponding to the sources in the mixture. Predictions are compared to the reference ground truth waveforms by searching over the space of (output, target) source order permutations, and we train using an objective function motivated by perceptual audio quality. We apply BioCPPNet to several species with unique vocal behavior, including macaques, bottlenose dolphins, and Egyptian fruit bats, and we evaluate reconstruction quality of separated waveforms using the scale-invariant signal-to-distortion ratio (SI-SDR) and downstream identity classification accuracy. We consider mixtures with two or three concurrent conspecific vocalizers, and we examine separation performance in open and closed speaker scenarios. To our knowledge, this paper redefines the state-of-the-art in end-to-end single-channel bioacoustic source separation in a permutation-invariant regime across a heterogeneous set of non-human species. This study serves as a major step toward the deployment of bioacoustic source separation systems for processing substantial volumes of previously unusable data containing overlapping bioacoustic signals.

Richard Kondratovich (1932–2017) – a scientist and rhododendron breeder

Richard Yazepovich Kondratovich was born in southeastern Latvia in the Saules Kalnes hamlet bordering Belarus. His father was the senior forester Yazep Kondratovich, and his mother Antonina was engaged in raising children, taking care of a big household and working part‑time in a local orphanage. In the period from 1939 to 1950 Richard went to school. From 1950 to 1955, he studied at the P. Stuchka Latvian State University (LSU). After graduating from the university, Richard Kondratovich was appointed director of the Botanical Garden of Latvian State University. In 1957, Richard started his scientific work on representatives of the genus Rhododendron. In 1964, he defended his Ph.D. thesis on the topic «Introduction of rhododendrons in the Latvian SSR». In 1981, two books by R. Kondratovich dedicated to rhododendrons were published in Russian. They aroused great inter‑est in this plant and have not lost their significance until now. In 1983, R. Kondratovich defended his doctoral thesis on the topic «Rhododendrons in the Latvian SSR». R. Kondratovich left the position of Director of the Botanical Garden in 1965 and later switched to teaching. Since 1957, R. Kondratovich has been carrying out research in rhododendron breeding, which expanded significantly after the opening of the «Babite» Rhododendron Breeding and Experimental Nursery. It covered an area of 12.1 hectares and was officially opened on July 1, 1980. In 2018, there were already 117 registered author’s varieties and 76 biological taxa (species and their forms) in the Nursery. The fruitful work was interrupted in February 2017 by illness. Richard Kondratovich passed away on February 17, 2017.

Animal Preferences: Implications of Sexual Selection Research for Empirical Aesthetics

The theory of sexual selection posits that sexual species make use of sensory preferences for physical traits and behavior displayed by potential sexual partners to inform mate choices. This suggests that the ability to compute hedonic liking for perceptual stimuli is widespread across biological taxa, have a long evolutionary history, and that the neurobiological processes involved in assigning liking to sensory objects may be conserved over this history. It remains unclear, though, to what degree specialized forms of hedonic evaluations found in humans, including aesthetic appreciation, share common computational principles with appraisals of sexual traits in other animals. Unfortunately, comparative work is very rare, and researchers in the fields of sexual selection and empirical aesthetics appear largely ignorant about each other’s work. In this paper I review recent important findings from sexual selection research in animals that directly have bearing on current discussions in empirical aesthetics about how to understanding the nature and uniqueness of human aesthetic appreciation. I argue that many of the outstanding issues in empirical aesthetics can only be resolved by comparing sensory evaluation in humans with other forms of sensory evaluation in other species.

Diversity and Antimicrobial Activity of Lichens-Associated Actinomycetes in Cibinong Science Centre (CSC) and Cibodas Botanical Garden (CBG) Indonesia

Bioprospecting has developed to all biological taxa including procaryotic. Actinomycetes become interesting procaryotic because of the ability to produce important secondary metabolite for human life. Actinomycetes are known as the largest antibiotic producer that has a broad range habitat. Some research has been done to find new antibiotic from the various habitat of actinomycetes. One of the interesting habitats of actinomycetes which never been explored in Indonesia is lichens... Lichens as the symbiotic structure of alga and fungi areknown as the ecological niche of various kinds of microorganisms including actinomycetes. Cibinong Science Centre (CSC) and Cibodas Botanical Garden (CBG) have various species of trees as the habitat of lichens. These areas are known as one of the research locations to explore the biodiversity of Indonesia. The aims of this research is to study the diversity and antimicrobial potency of actinomycetes isolated from 10 lichen samples with various type of thallus; crustose, fructose and foliose. Lichen samples were grown on the bark of 9 trees species in CSC and CBG. Isolation process used three agar media; HV, YIM6 and YIM711 with cycloheximide and nalidixic acid. Molecular identification based on 16S rRNA gene sequence. Antimicrobial activity was tested to 65 isolates by agar diffusion method to Bacillus subtilis BTCC B.612, Escherichia coli BTCC B.614, Candida albicans BTCC Y.33, Staphylococcus aureus BTCC B.611, Micrococcus luteus BTCC B.552. Isolation process retrieved 125 isolates with the highest number grow on HV agar medium. Based on the sample, the highest number of actinomycetes were isolated from crustose lichen attached on the bark of Averrhoea carambola. A total 69 isolates were identified as the genera Actinoplanes, Amycolatopsis, Angustibacter, Kribbella, Micromonospora, Mycobacterium, and Streptomyces. The screening process showed 24 isolates have antimicrobial activity, with the highest inhibitory activity against Micrococcus luteus BTCC B.552.

Single Fragment or Bulk Soil DNA Metabarcoding: Which is Better for Characterizing Biological Taxa Found in Surface Soils for Sample Separation?

In forensic geology casework, sample size typically limits routine characterization of material using bulk approaches. To address this, DNA-based characterization of biological taxa has received attention, as the taxa present can be useful for sample-to-sample comparisons and source attribution. In our initial work, low biodiversity was captured when DNA barcodes were Sanger-sequenced from plant and insect fragments isolated from 10 forensic-type surface soils. Considering some forensic laboratories now have access to massively parallel sequencing platforms, we assessed whether biological taxa present in the same surface soils could be better characterized using DNA metabarcoding. To achieve this, plant and animal barcodes were amplified and sequenced on an Illumina® MiniSeq for three different DNA sample types (n = 50): individual fragments used in our initial study, and 250 and 100 mg of bulk soil (from the 10 sites used in the initial study). A total of 572 unique target barcode sequences passed quality filtering and were used in downstream statistical analyses: 54, 321, and 285 for individual fragments, 100 mg, and 250 mg bulk soil samples, respectively. Plant barcodes permitted some spatial separation of sample sites in non-metric multidimensional scaling plots; better separation was obtained for samples prepared from bulk soil. This study confirmed that bulk soil DNA metabarcoding is a better approach for characterizing biological taxa present in surface soils, which could supplement traditional geologic examinations.

Linguistic Influence on Russian and American Ethnobiological Categorization

In the Russian language, nouns are classified by gender and animacy, whereas in English, nouns are not. Using triad-sorts of names for biological and non-biological taxa, a comparison of results provided by native speakers of both languages reveals that cognitive categorizations of animate and inanimate nomenclatural forms differ significantly between speakers of Russian and American English. Speakers of American English appear to categorize names for living nouns more by phenotype than do Russians, who in turn appear to classify nouns more frequently on the basis of linguistic features such as gender. These results are believed to be pertinent to the elicitation and construction of folk ethnobiology taxonomies.