Identifying wildlife observations on twitter

Despite the potential of social media for environmental monitoring, concerns remain about the quality and reliability of the information automatically extracted. Notably there are many observations of wildlife on Twitter, but their automated detection is a challenge due to the frequent use of wildlife related words in messages that have no connection with wildlife observation. We investigate whether and what type of supervised machine learning methods can be used to create a fully automated text classification model to identify genuine wildlife observations on Twitter, irrespective of species type or whether Tweets are geo-tagged. We perform experiments with various techniques for building feature vectors that serve as input to the classifiers, and consider how they affect classification performance. We compare three classification approaches and perform an analysis of the types of features that are indicative for genuine wildlife observations on Twitter. In particular, we compare some classical machine learning algorithms, widely used in ecology studies, with state-of-the-art neural network models. Results showed that the neural network-based model Bidirectional Encoder Representations from Transformers (BERT) outperformed the classical methods. Notably this was the case for a relatively small training corpus, consisting of less than 3000 instances. This reflects that fact that the BERT classifier uses a transfer learning approach that benefits from prior learning on a very much larger collection of generic text. BERT performed particularly well even for Tweets that employed specialised language relating to wildlife observations. The analysis of possible indicative features for wildlife Tweets revealed interesting trends in the usage of hashtags that are unrelated to official citizen science campaigns. The findings from this study facilitate more accurate identification of wildlife-related data on social media which can in turn be used for enriching citizen science data collections.     

Improving accuracy and efficiency in seagrass detection using state-of-the-art AI techniques

Seagrasses provide a wide range of ecosystem services in coastal marine environments. Despite their ecological and economic importance, these species are declining because of human impact. This decline has driven the need for monitoring and mapping to estimate the overall health and dynamics of seagrasses in coastal environments, often based on underwater images. However, seagrass detection from underwater digital images is not a trivial task; it requires taxonomic expertise and is time-consuming and expensive. Recently automatic approaches based on deep learning have revolutionised object detection performance in many computer vision applications, and there has been interest in applying this to automated seagrass detection from imagery. Deep learning–based techniques reduce the need for hardcore feature extraction by domain experts which is required in machine learning-based techniques. This study presents a YOLOv5-based one-stage detector and an EfficientDetD7–based two-stage detector for detecting seagrass, in this case, Halophila ovalis, one of the most widely distributed seagrass species. The EfficientDet-D7–based seagrass detector achieves the highest mAP of 0.484 on the ECUHO-2 dataset and mAP of 0.354 on the ECUHO-1 dataset, which are about 7% and 5% better than the state-of-the-art Halophila ovalis detection performance on those datasets, respectively. The proposed YOLOv5-based detector achieves an average inference time of 0.077 s and 0.043 s respectively which are much lower than the state-of-the-art approach on the same datasets.     

eThread: A Highly Optimized Machine Learning-Based Approach to Meta-Threading and the Modeling of Protein Tertiary Structures

Template-based modeling that employs various meta-threading techniques is currently the most accurate, and consequently the most commonly used, approach for protein structure prediction. Despite the evident progress in this field, accurate structure models cannot be constructed for a significant fraction of gene products, thus the development of new algorithms is required. Here, we describe the development, optimization and large-scale benchmarking of eThread, a highly accurate meta-threading procedure for the identification of structural templates and the construction of corresponding target-to-template alignments. eThread integrates ten state-of-the-art threading/fold recognition algorithms in a local environment and extensively uses various machine learning techniques to carry out fully automated template-based protein structure modeling. Tertiary structure prediction employs two protocols based on widely used modeling algorithms: Modeller and TASSER-Lite. As a part of eThread, we also developed eContact, which is a Bayesian classifier for the prediction of inter-residue contacts and eRank, which effectively ranks generated multiple protein models and provides reliable confidence estimates as structure quality assessment. Excluding closely related templates from the modeling process, eThread generates models, which are correct at the fold level, for >80% of the targets; 40–50% of the constructed models are of a very high quality, which would be considered accurate at the family level. Furthermore, in large-scale benchmarking, we compare the performance of eThread to several alternative methods commonly used in protein structure prediction. Finally, we estimate the upper bound for this type of approach and discuss the directions towards further improvements.     

Rapid-Viability PCR Method for Detection of Live, Virulent Bacillus anthracis in Environmental Samples

In the event of a biothreat agent release, hundreds of samples would need to be rapidly processed to characterize the extent of contamination and determine the efficacy of remediation activities. Current biological agent identification and viability determination methods are both labor- and time-intensive such that turnaround time for confirmed results is typically several days. In order to alleviate this issue, automated, high-throughput sample processing methods were developed in which real-time PCR analysis is conducted on samples before and after incubation. The method, referred to as rapid-viability (RV)-PCR, uses the change in cycle threshold after incubation to detect the presence of live organisms. In this article, we report a novel RV-PCR method for detection of live, virulent Bacillus anthracis, in which the incubation time was reduced from 14 h to 9 h, bringing the total turnaround time for results below 15 h. The method incorporates a magnetic bead-based DNA extraction and purification step prior to PCR analysis, as well as specific real-time PCR assays for the B. anthracis chromosome and pXO1 and pXO2 plasmids. A single laboratory verification of the optimized method applied to the detection of virulent B. anthracis in environmental samples was conducted and showed a detection level of 10 to 99 CFU/sample with both manual and automated RV-PCR methods in the presence of various challenges. Experiments exploring the relationship between the incubation time and the limit of detection suggest that the method could be further shortened by an additional 2 to 3 h for relatively clean samples.     

Instance segmentation and tracking of animals in wildlife videos: SWIFT - segmentation with filtering of tracklets

Reliably detecting and tracking animals in wildlife videos is an essential basis for researchers to be able to analyse animal behavior or recognize animal individuals. In order to correctly distinguish individual animals that are standing close to each other, bounding boxes around the animals are not sufficient. Instead, an exact contour of the animal, an instance mask, which is the result of an instance segmentation, is needed. In this paper, we present SWIFT, a novel multi-object tracking and segmentation (MOTS) pipeline that solves this task. We evaluate the functionality of our approach on a self-created wildlife video dataset containing red deer and fallow deer. Our dataset is one of the very few datasets in wildlife monitoring that is annotated with instance masks and tracking IDs. SWIFT significantly improves the quality of the instance masks compared to using a state-of-the-art instance segmentation approach from 0.432 average precision to 0.495 average precision. Our tracking algorithm uses multiple filtering steps to either delete tracks that are found incorrectly or to merge tracks that are not yet connected. This results in an increased multi-object tracking accuracy score in comparison to a state-of-the-art tracking approach from 57.2% to 63.8%, which means that our detected tracking results are less erroneous.     

Novel cost-efficient real-time PCR assays for detection and quantitation of Listeria monocytogenes

Listeriosis is a serious food-borne infection with mortality rates approaching 30%. Therefore, the rapid, cost-effective, and automated detection of Listeria monocytogenes throughout the food chain continues to be a major concern. Here we describe three novel quantitative real-time PCR assays for L. monocytogenes based on amplification of a target hlyA gene with SYBR Green I chemistry and hydrolysis probe (TaqMan MGB probe). In order to offer sensitive, rapid and robust tool of additional economical value the real-time PCR assays were designed and optimized to only 5 μl-reactions. All assays were evaluated by using different non-reference Listeria strains isolated from various food matrices. Results demonstrated specificity to L. monocytogenes with accurate quantification over a dynamic range of 5-6 log units with R² higher than 0.98 and amplification efficiencies reaching above 92%. The detection and quantification limits were as low as 165 genome equivalents. Comparison of novel assays to commercially available TaqMan® Listeria monocytogenes Detection Kit and previously published studies revealed similar specificity, sensitivity and efficiency, but greater robustness and especially cost-efficiency in the view of smaller reaction volumes and continuous increase in sample throughput.     

New technologies in the mix: Assessing N‐mixture models for abundance estimation using automated detection data from drone surveys

1. Reliable estimates of abundance are critical in effectively managing threatened species, but the feasibility of integrating data from wildlife surveys completed using advanced technologies such as remotely piloted aircraft systems (RPAS) and machine learning into abundance estimation methods such as N‐mixture modeling is largely unknown due to the unique sources of detection errors associated with these technologies.
2. We evaluated two modeling approaches for estimating the abundance of koalas detected automatically in RPAS imagery: (a) a generalized N‐mixture model and (b) a modified Horvitz–Thompson (H‐T) estimator method combining generalized linear models and generalized additive models for overall probability of detection, false detection, and duplicate detection. The final estimates from each model were compared to the true number of koalas present as determined by telemetry‐assisted ground surveys.
3. The modified H‐T estimator approach performed best, with the true count of koalas captured within the 95% confidence intervals around the abundance estimates in all 4 surveys in the testing dataset (n = 138 detected objects), a particularly strong result given the difficulty in attaining accuracy found with previous methods.
4. The results suggested that N‐mixture models in their current form may not be the most appropriate approach to estimating the abundance of wildlife detected in RPAS surveys with automated detection, and accurate estimates could be made with approaches that account for spurious detections.

     

Real-time observation,early warning and forecasting phytoplankton blooms by integrating in situ automated online sondes and hybrid evolutionary algorithms

Phytoplankton bloom is one of the most serious threats to water resource, and remains a global challenge in environmental management. Real-time monitoring and forecasting the dynamics of phytoplankton and early warning the risks are critical steps in an effective environmental management. Automated online sondes have been widely used for in situ real-time monitoring of water quality due to their high reliability and low cost. However, the knowledge of using real-time data from those sondes to forecast phytoplankton blooms has been seldom addressed. Here we present an integrated system for real-time observation, early warning and forecasting of phytoplankton blooms by integrating automated online sondes and the ecological model. Specifically, based on the high-frequency data from automated online sondes in Xiangxi Bay of Three Gorges Reservoir, we successfully developed 1–4 days ahead forecasting models for chlorophyll a (chl a) concentration with hybrid evolutionary algorithms (HEAs). With the predicted concentration of chl a, we achieved a high precision in 1–7 days ahead early warning of good (chl a < 25 μg/L) and eutrophic (chl a 8–25 μg/L) conditions; however only achieved an acceptable precision in 1–2 days ahead early warning of hypertrophic condition (chl a ≥ 25 μg/L). Our study shows that the optimized HEAs achieved an acceptable performance in real-time short-term forecasting and early warning of phytoplankton blooms with the data from the automated in situ sondes. This system provides an efficient way in real-time monitoring and early warning of phytoplankton blooms, and may have a wide application in eutrophication monitoring and management.     

Surveying an endangered saproxylic beetle, Osmoderma eremita, in Mediterranean woodlands: a comparison between different capture methods

Measuring population size is riddled with difficulties for wildlife biologists and managers, and in the case of rare species, it is sometimes practically impossible to estimate abundance, whereas estimation of occupancy is possible. Furthermore, obtaining reliable population size estimates is not straightforward, as different sampling techniques can give misleading results. A mark-recapture study of the endangered saproxylic beetle Osmoderma eremita was performed in central Italy by applying four independent capture methods within a study area where 116 hollow trees were randomly selected to set traps. Detection probability and population size estimates were drawn from each of these four capture methods. There were strong differences in detection probability among methods. Despite using pheromone and beetle manipulation, capture histories were not affected by trap-happiness or trap-shyness. Population size estimates varied considerably in both abundance and precision by capture method. A number of 0.5 and 0.2 adult beetles per tree was estimated using the whole data set by closed and open population models, respectively. Pitfall trap appeared the optimal method to detect the occurrence of this species. Since in the southern part of its distribution range, a single population of O. eremita is widespread in the landscape, and includes beetles from more than one hollow tree, conservation efforts should focus not only on preserving few and isolated monumental hollow trees, but should be extended to large stands.     

Faster, Simpler, More-Specific Methods for Improved Molecular Detection of Phytophthora ramorum in the Field

Phytophthora ramorum is the causal agent of sudden oak death. The pathogen also affects a wide range of tree, shrub, and herbaceous species in natural and landscaped environments as well as plants in the nursery industry. A TaqMan real-time PCR method for the detection of this pathogen in the field has been described previously; this paper describes the development of a number of assays based on this method which have various advantages for use in the field. A scorpion real-time PCR assay that is twice as fast as TaqMan was developed, allowing the detection of P. ramorum in less than 30 min. Also designed was a loop-mediated isothermal amplification (LAMP) assay, which allowed sensitive and specific detection of P. ramorum in 45 min using only a heated block. A positive reaction was identified by the detection of the LAMP product by color change visible to the naked eye.     

Detection of Bacterial 16S rRNA and Identification of Four Clinically Important Bacteria by Real-Time PCR

Within the paradigm of clinical infectious disease research, Acinetobacter baumannii, Escherichia coli, Klebsiella pneumoniae, and Pseudomonas aeruginosa represent the four most clinically relevant, and hence most extensively studied bacteria. Current culture-based methods for identifying these organisms are slow and cumbersome, and there is increasing need for more rapid and accurate molecular detection methods. Using bioinformatic tools, 962,279 bacterial 16S rRNA gene sequences were aligned, and regions of homology were selected to generate a set of real-time PCR primers that target 93.6% of all bacterial 16S rRNA sequences published to date. A set of four species-specific real-time PCR primer pairs were also designed, capable of detecting less than 100 genome copies of A. baumannii, E. coli, K. pneumoniae, and P. aeruginosa. All primers were tested for specificity in vitro against 50 species of Gram-positive and –negative bacteria. Additionally, the species-specific primers were tested against a panel of 200 clinical isolates of each species, randomly selected from a large repository of clinical isolates from diverse areas and sources. A comparison of culture and real-time PCR demonstrated 100% concordance. The primers were incorporated into a rapid assay capable of positive identification from plate or broth cultures in less than 90 minutes. Furthermore, our data demonstrate that current targets, such as the uidA gene in E.coli, are not suitable as species-specific genes due to sequence variation. The assay described herein is rapid, cost-effective and accurate, and can be easily incorporated into any research laboratory capable of real-time PCR.     

Optimizing detection of the bobwhite reproduction call using passive acoustic monitoring

Northern bobwhite (Colinus virginianus) populations have declined across much of their range. In response to these declines, wildlife biologists and managers have increased survey efforts and tried to optimize detection and capitalize on technological advances to improve population estimates and cost-effectiveness. Our objective was to determine how environmental conditions influence detection of the reproduction call, or whistle, of masked bobwhite (C. v. ridgwayi), an endangered subspecies of northern bobwhite, using autonomous recording units (ARUs). We estimated the call intensity of the masked bobwhite reproduction call as 112 ± 0.5 decibels (mean ± SE) at 10 cm. We then broadcasted 16,284 calls during 17 trials to compare manual and automated call detection in recordings collected with ARUs. We used these data to model detectability of a bobwhite reproduction call, for when the bird is present and available, as a function of distance and weather conditions using generalized linear mixed models with trial as a random effect. Regardless of detection type, one model structure was competitive and suggested detection probability was a function of distance, wind speed, and wind direction. Detectability decreased with increased distance and wind speed and was influenced by wind direction. We demonstrate the use of our results to predict the probability of detecting a reproduction call during ARU-based monitoring efforts. By understanding the effects of environmental factors on the detection of a bobwhite reproductive call, bobwhite surveys can be improved.     

PCR-based molecular diagnosis of botulism (types C and D) outbreaks in aquatic birds

Clostridium botulinum is a strictly anaerobic spore forming bacterium found in soil and sediments, producing botulinum neurotoxins. Group III strains of this organism are only able to produce neurotoxin types C, mosaic C/D, D/C, and D, which are associated with bird botulism. The threats and outbreak cycle of bird botulism are enhanced in the aquatic environment via a food web-environment-avian interface in response to global climatic changes. The aim of this review was to describe and discuss the development of PCR-based markers, diagnostic assays, and applications with special emphasis to botulism detection in aquatic birds. We employed a text-mining approach for collection of current diagnostic information to bird botulism in aquatic environments. Using the PubMed search engine, we have comprehensively collected relevant information from 124 articles and then summated for the purpose of this review. Genes coding for botulinum neurotoxin (bont/C, bont/C/D, bont/D/C, and bont/D), nontoxic components (ha70, ha17, ha33, and ntnh), and flagellin (fliC) were the molecular markers most commonly found for the genotyping of group III strains from environmental samples. “GeneDisc” real-time PCR system was a robust and reliable diagnostic technique for discrimination of neurotoxin types in the large sampled areas. PCR-based assays were the most perceptive and widely established for detection of bird botulism, even if several biochemical and molecular diagnostic techniques have been available. Thus, timely and accurate identification of its mortality is needed to provide a biosecurity, disease control management, and conservation of aquatic birds. PCR-based diagnosis is a promising alternative to the mouse bioassay. Hence, the present paper makes an attempt to review the PCR-based detection assays including primers, specificity, sensitivity, and detection limit and explore their potential applications in wildlife microbiology.     

Atrial flutter and atrial tachycardia detection using Bayesian approach with high resolution time–frequency spectrum from ECG recordings

Contemporary methods of atrial flutter (AFL), atrial tachycardia (AT), and atrial fibrillation (AF) monitoring, although superior to the standard 12-lead ECG and symptom-based monitoring, are unable to accurately discriminate between AF, AFL and AT. Thus, there is a need to develop accurate, automated, and comprehensive atrial arrhythmia detection algorithms using standard ECG recorders. To this end, we have developed a sensitive and real-time realizable algorithm for accurate AFL and AT detection using any standard electrocardiographic recording. Our novel method for automatic detection of atrial flutter and atrial tachycardia uses a Bayesian approach followed by a high resolution time–frequency spectrum. We find the TQ interval of the electrocardiogram (ECG) corresponding to atrial activity by using a particle filter (PF), and analyze the atrial activity with a high resolution time–frequency spectral method: variable frequency complex demodulation (VFCDM). The rationale for using a high-resolution time–frequency algorithm is that our approach tracks the time-varying fundamental frequency of atrial activity, where AT is within 2.0–4.0 Hz, AFL is within 4.0–5.3 Hz and NSR is found at frequencies less than 2.0 Hz. For classifications of AFL (n = 22), AT (n = 10) and normal sinus rhythms (NSR) (n = 29), we found that our approach resulted in accuracies of 0.89, 0.87 and 0.91, respectively; the overall accuracy was 0.88.     

Global Positioning System Data-Loggers: A Tool to Quantify Fine-Scale Movement of Domestic Animals to Evaluate Potential for Zoonotic Transmission to an Endangered Wildlife Population

Domesticated animals are an important source of pathogens to endangered wildlife populations, especially when anthropogenic activities increase their overlap with humans and wildlife. Recent work in Tanzania reports the introduction of Cryptosporidium into wild chimpanzee populations and the increased risk of ape mortality associated with SIVcpz-Cryptosporidium co-infection. Here we describe the application of novel GPS technology to track the mobility of domesticated animals (27 goats, 2 sheep and 8 dogs) with the goal of identifying potential routes for Cryptosporidium introduction into Gombe National Park. Only goats (5/27) and sheep (2/2) were positive for Cryptosporidium. Analysis of GPS tracks indicated that a crop field frequented by both chimpanzees and domesticated animals was a potential hotspot for Cryptosporidium transmission. This study demonstrates the applicability of GPS data-loggers in studies of fine-scale mobility of animals and suggests that domesticated animal–wildlife overlap should be considered beyond protected boundaries for long-term conservation strategies.     

Exploiting single-molecule transcript sequencing for eukaryotic gene prediction

We develop a method to predict and validate gene models using PacBio single-molecule, real-time (SMRT) cDNA reads. Ninety-eight percent of full-insert SMRT reads span complete open reading frames. Gene model validation using SMRT reads is developed as automated process. Optimized training and prediction settings and mRNA-seq noise reduction of assisting Illumina reads results in increased gene prediction sensitivity and precision. Additionally, we present an improved gene set for sugar beet (Beta vulgaris) and the first genome-wide gene set for spinach (Spinacia oleracea). The workflow and guidelines are a valuable resource to obtain comprehensive gene sets for newly sequenced genomes of non-model eukaryotes.

Electronic supplementary material

The online version of this article (doi:10.1186/s13059-015-0729-7) contains supplementary material, which is available to authorized users.     

Lateral-Flow Device for Diagnosis of Fungal Infection

Invasive aspergillosis (IA) is a life-threatening complication of haematological malignancy and haematopoietic stem cell transplantation caused by the ubiquitous fungus Aspergillus. Current diagnosis of IA is multifaceted relying on data from clinical, radiological, and microbiological sources. The detection of Aspergillus biomarkers provides strategies both to pre-empt and to exclude disease, but the choice of biomarker assays is limited, requiring specialist equipment and training. This review examines recent advances in the accurate diagnosis of IA through the development of an Aspergillus lateral-flow device (LFD) incorporating a monoclonal antibody, JF5, which detects an antigenic marker of active infection. Recent trials using bronchoalveolar lavage fluids and serum samples from humans and from animal models of disease, have demonstrated the utility of the LFD as a rapid and user-friendly adjunct test for the quick and accurate diagnosis of pulmonary infections.     

A Quantitative Real-time PCR Assay for Quantification of Viable Listeria Monocytogenes Cells After Bacteriocin Injury in Food-First Insights

Quantitative real-time PCR may be a rapid and automated procedure for detection of bacterial pathogens from food samples. Nevertheless, when testing the effects of antimicrobials on the viability of bacterial pathogens in foods, we found that DNA from dead cells interfered greatly in the detection of viable Listeria monocytogenes after treatment with the broad-spectrum bacteriocin enterocin AS-48. To overcome this problem, a quantitative real-time PCR (qRT-PCR) assay based on bacterial mRNA was adapted to quantify viable L. monocytogenes in food after bacteriocin treatments. The procedure allowed a better and faster estimation of viable cells compared to PALCAM viable cell counts when the threshold level was 2 log units/g of food, while PALCAM viable count allowed detection of one log unit/g. This procedure may be useful to verify the efficacy of bacteriocins against L. monocytogenes in foods.     

Surface Immuno-Functionalisation for the Capture and Detection of Vibrio Species in the Marine Environment: A New Management Tool for Industrial Facilities

Bacteria from the genus Vibrio are a common and environmentally important group of bacteria within coastal environments and include species pathogenic to aquaculture organisms. Their distribution and abundance are linked to specific environmental parameters, including temperature, salinity and nutrient enrichment. Accurate and efficient detection of Vibrios in environmental samples provides a potential important indicator of overall ecosystem health while also allowing rapid management responses for species pathogenic to humans or species implicated in disease of economically important aquacultured fish and invertebrates. In this study, we developed a surface immuno-functionalisation protocol, based on an avidin-biotin type covalent binding strategy, allowing specific sandwich-type detection of bacteria from the Vibrio genus. The assay was optimized on 12 diverse Vibrio strains, including species that have implications for aquaculture industries, reaching detection limits between 7×10³ to 3×10⁴ cells mL⁻¹. Current techniques for the detection of total Vibrios rely on laborious or inefficient analyses resulting in delayed management decisions. This work represents a novel approach for a rapid, accurate, sensitive and robust tool for quantifying Vibrios directly in industrial systems and in the environment, thereby facilitating rapid management responses.