INDIVIDUAL IDENTIFICATION: BIRDS
Sherley R. B., Burghardt T., Barham P. J., Campbell N., Cuthill I. C. (2010): Spotting the difference: towards fully-automated population monitoring of African penguins Spheniscus demersus. Endangered Species Research 11: 101-111.
FULL TEXT
Abstract
Placing external monitoring devices onto seabirds can have deleterious effects on welfare and performance, and even the most benign marking and identification methods return sparse population data at a huge time and effort cost. Consequently, there is growing interest in methods that minimise disturbance but still allow robust population monitoring. We have developed a computer vision system that automatically creates a unique biometric identifier for individual adult African penguins Spheniscus demersus using natural markings in the chest plumage and matches this against a population database. We tested this non-invasive system in the field at Robben Island, South Africa. False individual identifications of detected penguins occurred in less than 1 in 10000 comparisons (n = 73600, genuine acceptance rate = 96.7%) to known individuals. The monitoring capacity in the field was estimated to be above 13% of the birds that passed a camera (n = 1453). A significant increase in this lower bound was recorded under favourable conditions. We conclude that the system is suitable for population monitoring of this species: the demonstrated sensitivity is comparable to computer-aided animal biometric monitoring systems in the literature. A full deployment of the system would identify more penguins than is possible with a complete exploitation of the current levels of flipper banding at Robben Island. Our study illustrates the potential of fully-automated, non-invasive, complete population monitoring of wild animals.
Murn C. (2012): Field identification of individual White-headed Vultures Trigonoceps occipitalis using plumage patterns – an information theoretic approach. Bird Study 59: 515-521.
FULL TEXT
Abstract
To highlight the distinctiveness of the median wing covert pattern in White-headed Vultures and demonstrate the reliably high information content contained in this pattern. Photographs of 30 wild White-headed Vultures were image-processed and overlaid with an analysis grid. An information theoretic approach was used to determine the probability of a specific median wing covert pattern recurring in the population. This probability determines the information content of each pattern. The information content of median wing covert patterns is high (median content 23.54 bits) and the probability of pattern recurrence in a population of 10 000 birds is low (P = 2.04 × 10−3). The likelihood of the pattern changing over time is low. White-headed Vultures show variation in their median wing covert pattern that is sufficient for birds to be individually identifiable in the field. This non-invasive identification technique is reliable and is suitable for cataloguing local and regional populations of adult White-headed Vultures, thus facilitating mark–recapture studies or other studies that require identification of individuals.
Williams E. R., Thomson B. (2015): Improving population estimates of Glossy Black-Cockatoos (Calyptorhynchus lathami) using photo-identification. Emu-Austral Ornithology 115: 360-367.
FULL TEXT
Abstract
Site-based population estimates of the threatened Glossy Black-Cockatoo (‘GBC’; Calyptorhynchus lathami) are often calculated based on age and sex details from transect counts. However, these estimates do not distinguish individual birds, which may result in over-or under-estimation of the population. Two methods were used to estimate GBC populations in Mugii Murum-ban State Conservation Area, New South Wales: (1) the traditional transect method, and (2) a photographic method, which used plumage patterns (primarily the yellow facial feathers of females) and other supporting features to discriminate between foraging GBC family units. The second method has been used previously on Kangaroo Island, South Australia. A catalogue with a matrix of discriminating features was established based on the photographic method; this resulted in a higher population estimate than the transect method in two seasons (winter and spring), as well as providing an annual population estimate, and information on breeding dynamics and local movement of individual family units between foraging habitat. Recommendations for the application of the photographic method are provided. The method provides benefits at both the local scale (with more accurate site population estimates and information on population dynamics) and, with widespread adoption and national cataloguing, valuable knowledge on regional movement patterns and distribution.
Hoy S. R., Ball R. E., Lambin X., Whitfield D. P., Marquiss M. (2016): Genetic markers validate using the natural phenotypic characteristics of shed feathers to identify individual northern goshawks Accipiter gentilis. Journal of Avian Biology 47: 443-447.
FULL TEXT
Abstract
The recognition of individual animals is essential for many types of ecological research, as it enables estimates of demographic parameters such as population size, survival and reproductive rates. A popular method of visually identifying individuals uses natural variations in spot, stripe or scar markings. Although several studies have assessed the accuracy of these methods in mammals, crustaceans and fish, there have been few attempts to determine whether phenotypic characteristics are accurate when used for birds. Furthermore, even less is known about whether shed or moulted body parts can be reliably used to visually identify individuals. Here we assessed the accuracy of using phenotypic characteristics to identify avian individuals using a double‐marking experiment, whereby nine microsatellite genetic markers and natural markings on shed feathers were used to independently identify northern goshawks Accipiter gentilis. Phenotypic and genetic identification of individuals was consistent in 94.4% (51/54) comparisons. Our results suggest that the phenotypic characteristics of shed feathers can be reliably used as a non‐invasive and relatively inexpensive technique to monitor populations of an elusive species, the northern goshawk, without having to physically re‐capture or re‐sight individuals. We posit that using natural markings on shed feathers will also be a reliable method of identifying individuals in avian species with similar phenotypic characteristics, such as other Accipiter species.
Lehtonen P. J., Lappalainen J. (2017): Individual identification of Black-throated Divers (Gavia arctica). Ornis Fennica 94: 2-12.
FULL TEXT
Abstract
The potential to identify individual Black-throated Divers (Gavia arctica) on the basis of breeding plumage features was explored using 278 photos, including two paired birds followed during the years 2007-2015 at a specific breeding location. Observations were focused on: 1) white lines on the sides of neck, 2) mantle having rows of sharply contrasting white squares, and 3) small white spots on lesser and median coverts. In photos, the number of white lines on the sides of neck varied from four to seven (mean = 5.0, n = 278), and the second line from the head was the highest in 92.1% of the photos. The number of “white square” rows on the mantle varied from 11 to 14 and the small white spots on coverts from 27 to 67. Identification of individual Black-throated Divers was potentially easiest if the plumage had some special patterns (19.4% of birds, n = 278). Plumage remained the same in the followed pair between years, as was also shown by the discriminant analysis, since the followed pair was correctly classified by sex but not by sides showing that sides are similar. To estimate whether it is possible to separate these two birds from other birds, a second discriminant analysis was accomplished. Thus, 125 other birds were added to analysis as a third group together with the followed pair (female and male, nine years and n = 18 per sex). The linear discriminant analysis yielded a classification rate of 70.8% in original analysis and 69.6% based on the leave-one-out analysis (n = 161). These analyses were based on the relative height of the neck lines, their average relative height and standard deviation. When the number of white spots were added to this discriminant analysis, a correct classification rate of 77.4% in original analysis and 75.7% in the leave-one-out analysis was obtained (n = 115). These following nesting pairs during their breeding seasons in different years. Presumed female and male Black-throated Diver could be distinguished based on the shape of the forehead.
Méndez D., Marsden S., Lloyd H. (2019): Assessing population size and structure for Andean Condor Vultur gryphus in Bolivia using a photographic ‘capture‐recapture’ method. Ibis 161: 867-877.
FULL TEXT
Abstract
The Andean Condor Vultur gryphus is a globally threatened and declining species. Problems of surveying Andean Condor populations using traditional survey methods are particularly acute in Bolivia, largely because only few roosts are known there. However, similar to other vulture species, Andean Condors aggregate at animal carcasses, and are individually recognizable due to unique morphological characteristics (size and shape of male crests and pattern of wing coloration). This provided us with an opportunity to use a capture-recapture (‘sighting-resighting’) modelling framework to estimate the size and structure of an Andean Condor population in Bolivia using photographs of individuals taken at observer-established feeding stations. Between July and December 2014, 28 feeding stations were established in five different zones throughout the eastern Andean region of Bolivia, where perched and flying Andean Condors were photographed. Between one and 57 (mean = 20.2 ± 14.6 sd) Andean Condors were recorded visiting each feeding station and we were able to identify 456 different individuals, comprising 134 adult males, 40 sub-adult males, 79 juvenile males, 80 adult females, 30 sub-adult females and 93 juvenile females. Open population capture-recapture models produced population estimates ranging from 52 ± 14 (se) individuals to 678 ± 269 individuals across the five zones, giving a total of 1388 ± 413 sd individuals, which is roughly 20% of the estimated Andean Condor global population. Future trials of this method need to consider explicitly knowledge of Andean Condor movements and home-ranges, habitat preferences when selecting suitable sites as feeding stations, juvenile movements and other behaviours. Sighting-resighting methods have considerable potential to increase the accuracy of surveys of Andean Condors and other bird species with unique individual morphological characteristics.
Stowell D., Petrusková T., Šálek M., Linhart P. (2019): Automatic acoustic identification of individuals in multiple species: improving identification across recording conditions. Journal of the Royal Society Interface 16: 20180940.
FULL TEXT
Abstract
Many animals emit vocal sounds which, independently from the sounds’ function, contain some individually distinctive signature. Thus the automatic recognition of individuals by sound is a potentially powerful tool for zoology and ecology research and practical monitoring. Here, we present a general automatic identification method that can work across multiple animal species with various levels of complexity in their communication systems. We further introduce new analysis techniques based on dataset manipulations that can evaluate the robustness and generality of a classifier. By using these techniques, we confirmed the presence of experimental confounds in situations resembling those from past studies. We introduce data manipulations that can reduce the impact of these confounds, compatible with any classifier. We suggest that assessment of confounds should become a standard part of future studies to ensure they do not report over-optimistic results. We provide annotated recordings used for analyses along with this study and we call for dataset sharing to be a common practice to enhance the development of methods and comparisons of results.
Chen G., Xia C., Zhang Y. (2020): Individual identification of birds with complex songs: The case of green-backed flycatchers Ficedula elisae. Behavioural Processes 173: 104063.
FULL TEXT
Abstract
Vocal individual identification has been demonstrated in many animals, with discriminant function analysis (DFA) and spectrographic cross-correlation (SPCC) being the two most frequent methods. Successful vocal individual identification requires high among-individual differences and within-individual stability over time for vocal features. Lack of vocal individual identification is common in songbirds with complex songs, and most vocal individual identification studies are made in bird species with simple vocalizations. Here, we applied vocal individual identification with the two methods on a songbird, green-backed flycatcher Ficedula elisae. We based its complex songs by division into first, second, and third phrases. DFA resulted in a correct distinction rate of 94.5 % between one first-phrase type and another. SPCC similarity was significantly higher within than among types for first and second phrases, respectively. For first-phrase types with recordings from different days during a breeding season, the correct DFA rate was 87.1 %. SPCC similarity within type changed significantly among days, but was still significantly higher than that among types. In conclusion, first phrases of the complex songs met the two requirements and could be effectively used for vocal individual identification in this species. This study filled a gap in vocal individual identification in birds with complex songs.
Ferreira A. C., Silva L. R., Renna F., Brandl H. B., Renoult J. P., Farine D. R., Covas R., Doutrelant C. (2020): Deep learning‐based methods for individual recognition in small birds. Methods in Ecology and Evolution 11: 1072-1085.
FULL TEXT
Abstract
Individual identification is a crucial step to answer many questions in evolutionary biology and is mostly performed by marking animals with tags. Such methods are well-established, but often make data collection and analyses time-consuming, or limit the contexts in which data can be collected. Recent computational advances, specifically deep learning, can help overcome the limitations of collecting large-scale data across contexts. However, one of the bottlenecks preventing the application of deep learning for individual identification is the need to collect and identify hundreds to thousands of individually labelled pictures to train convolutional neural networks (CNNs). Here we describe procedures for automating the collection of training data, generating training datasets, and training CNNs to allow identification of individual birds. We apply our procedures to three small bird species, the sociable weaver Philetairus socius, the great tit Parus major and the zebra finch Taeniopygia guttata, representing both wild and captive contexts. We first show how the collection of individually labelled images can be automated, allowing the construction of training datasets consisting of hundreds of images per individual. Second, we describe how to train a CNN to uniquely re-identify each individual in new images. Third, we illustrate the general applicability of CNNs for studies in animal biology by showing that trained CNNs can re-identify individual birds in images collected in contexts that differ from the ones originally used to train the CNNs. Finally, we present a potential solution to solve the issues of new incoming individuals. Overall, our work demonstrates the feasibility of applying state-of-the-art deep learning tools for individual identification of birds, both in the laboratory and in the wild. These techniques are made possible by our approaches that allow efficient collection of training data. The ability to conduct individual recognition of birds without requiring external markers that can be visually identified by human observers represents a major advance over current methods.
Šulc M., Hughes A. E., Troscianko J., Štětková G., Procháazka P., Požgayová M., Piálek L., Piálkova R, Brlík V., Honza M. (2021): Automatic identification of bird females using egg phenotype. Zoological Journal of the Linnean Society 2021: zlab051.
FULL TEXT
Abstract
Individual identification is crucial for studying animal ecology and evolution. In birds this is often achieved by capturing and tagging. However, these methods are insufficient for identifying individuals/species that are secretive or difficult to catch. Here, we employ an automatic analytical approach to predict the identity of bird females based on the appearance of their eggs, using the common cuckoo (Cuculus canorus) as a model species. We analysed 192 cuckoo eggs using digital photography and spectrometry. Cuckoo females were identified from genetic sampling of nestlings, allowing us to determine the accuracy of automatic (unsupervised and supervised) and human assignment. Finally, we used a novel analytical approach to identify eggs that were not genetically analysed. Our results show that individual cuckoo females lay eggs with a relatively constant appearance and that eggs laid by more genetically distant females differ more in colour. Unsupervised clustering had similar cluster accuracy to experienced human observers, but supervised methods were able to outperform humans. Our novel method reliably assigned a relatively high number of eggs without genetic data to their mothers. Therefore, this is a cost-effective and minimally invasive method for increasing sample sizes, which may facilitate research on brood parasites and other avian species.
Martin K., Adam O., Obin N., Dufour V. (2022): Rookognise: Acoustic detection and identification of individual rooks in field recordings using multi-task neural networks. Ecological Informatics 72: 101818.
FULL TEXT
Abstract
Individual-level monitoring is essential in many behavioural and bioacoustics studies. Collecting and annotating those data is costly in terms of human effort, but necessary prior to conducting analysis. In particular, many studies on bird vocalisations also involve manipulating the animals or human presence during observations, which may bias vocal production. Autonomous recording units can be used to collect large amounts of data without human supervision, largely removing those sources of bias. Deep learning can further facilitate the annotation of large amounts of data, for instance to detect vocalisations, identify the species, or recognise the vocalisation types in recordings. Acoustic individual identification, however, has so far largely remained limited to a single vocalisation type for a given species. This has limited the use of those techniques for automated data collection on raw recordings, where many individuals can produce vocalisations of varying complexity, potentially overlapping one another, with the additional presence of unknown and varying background noise. This paper aims at bridging this gap by developing a system to identify individual animals in those difficult conditions. Our system leverages a combination of multi-scale information integration, multi-channel audio and multi-task learning. The multi-task learning paradigm is based the overall task into four sub-tasks, three of which are auxiliary tasks: the detection and segmentation of vocalisations against other noises, the classification of individuals vocalising at any point during a sample, and the sexing of detected vocalisations. The fourth task is the overall identification of individuals. To test our approach, we recorded a captive group of rooks, a Eurasian social corvid with a diverse vocal repertoire. We used a multi-microphone array and collected a large scale dataset of time-stamped and identified vocalisations recorded, and found the system to work reliably for the defined tasks. To our knowledge, the system is the first to acoustically identify individuals regardless of the vocalisation produced. Our system can readily assist data collection and individual monitoring of groups of animals in both outdoor and indoor settings, even across long periods of time, and regardless of a species’ vocal complexity. All data and code used in this article is available online.