Roman Gula
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Roman Gula has a total of 179 co-authors. Here are the top co-authors by number of shared publications:
- Jörn Theuerkauf
Museum and Institute of Zoology, Polish Academy of Sciences, Warsaw, Poland
54 shared publications - Ralph Kuehn
Molecular Zoology, Department of Zoology, TUM School of Life Sciences, Technical University of Munich, Munich, Germany
17 shared publications Keisuke Ueda
Rikkyo University Tôkyô Japan
11 shared publicationsYuji Okahisa
Rikkyo University Tôkyô Japan
11 shared publications- Alfredo Attisano
Museum and Institute of Zoology, Polish Academy of Sciences, Warsaw, Poland
9 shared publications Nozomu J. Sato
Japan Bird Research Association Tokyo Japan
8 shared publications
Latest Publications
- Open Access
- 23 May 2026
- Published by Elsevier in Ecological Informatics
- Vol. 96
From relative to metric: Calibrating AI-based monocular depth learning models for distance sampling in wildlife monitoring applications
- 23 May 2026
- Published by Elsevier in Ecological Informatics
- Vol. 96
Open AccessFrom relative to metric: Calibrating AI-based monocular depth learning models for distance sampling in wildlife monitoring applications
- Open Access
- 23 October 2025
- Published by Wiley in Journal of Animal Ecology
- Vol. 94 (12) , 2597-2614
An integrative, peer‐reviewed and open‐source cooperative‐breeding database (Co‐ BreeD )
Large‐scale, cross‐species comparative analyses on cooperative breeding—where individuals care for the offspring of other group members—are important for understanding sociality and cooperation. However, the datasets that facilitate these analyses are often limited in precision. To advance comparative research on cooperative breeding, we hereby introduce the Cooperative‐Breeding Database (Co‐BreeD) for birds and mammals. We describe key features of Co‐BreeD's structure: (i) integration of complementary datasets, each presenting a biological parameter relevant to cooperative‐breeding research; (ii) sample‐based (i.e. multiple samples per species linked to an exact sampling location and period); and (iii) open‐source. Respectively, these features enable: (a) comprehensive identification of cooperative‐breeding species according to the user's chosen definition, (b) linking intra‐ and inter‐specific variation in traits with fine‐scale environmental parameters and (c) enabling the research community to correct and expand this database. We present the initial Co‐BreeD dataset, which estimates the prevalence of breeding events involving potential alloparents in 460 populations of 324 species, including 6 human populations (No. total = 43,247 breeding events). We conclude by demonstrating: (i) how Co‐BreeD can improve comparative research (e.g. by enabling the study of cooperative breeding as a continuous rather than a binary trait); and (ii) that cooperative breeding is probably more prevalent than previously estimated in birds and mammals.
- 23 October 2025
- Published by Wiley in Journal of Animal Ecology
- Vol. 94 (12) , 2597-2614
Open AccessAn integrative, peer‐reviewed and open‐source cooperative‐breeding database (Co‐ BreeD )
Large‐scale, cross‐species comparative analyses on cooperative breeding—where individuals care for the offspring of other group members—are important for understanding sociality and cooperation. However, the datasets that facilitate these analyses are often limited in precision. To advance comparative research on cooperative breeding, we hereby introduce the Cooperative‐Breeding Database (Co‐BreeD) for birds and mammals. We describe key features of Co‐BreeD's structure: (i) integration of complementary datasets, each presenting a biological parameter relevant to cooperative‐breeding research; (ii) sample‐based (i.e. multiple samples per species linked to an exact sampling location and period); and (iii) open‐source. Respectively, these features enable: (a) comprehensive identification of cooperative‐breeding species according to the user's chosen definition, (b) linking intra‐ and inter‐specific variation in traits with fine‐scale environmental parameters and (c) enabling the research community to correct and expand this database. We present the initial Co‐BreeD dataset, which estimates the prevalence of breeding events involving potential alloparents in 460 populations of 324 species, including 6 human populations (No. total = 43,247 breeding events). We conclude by demonstrating: (i) how Co‐BreeD can improve comparative research (e.g. by enabling the study of cooperative breeding as a continuous rather than a binary trait); and (ii) that cooperative breeding is probably more prevalent than previously estimated in birds and mammals.
- Open Access
- 20 August 2025
- Published by Springer Nature in Journal of Ornithology
- Vol. 167 (1) , 267-279
Density distribution of the charismatic Kagu to guide conservation of endangered endemic rainforest species in New Caledonia
The Kagu ( Rhynochetos jubatus ) is an endangered endemic bird and emblem of New Caledonia, thus an ideal flagship species for conservation management and planning of protected areas on Grande Terre, the main island. We assessed the density distribution of Kagu by combining results of analyses at two spatial scales. At the local scale, we estimated density in two protected Kagu strongholds. At the island-wide scale, we modelled distribution using MaxEnt over the entire Kagu range on Grande Terre. We then combined these approaches to convert distribution probability into a density distribution. Proportion of rainforest and size of the largest rainforest patch within 250 m × 250 m raster cells were the main factors predicting Kagu distribution. Another important factor was distance to settlements, which is likely related to dog presence, the only significant predator of Kagu. The model identified nearly 2000 km 2 as habitat, which could potentially support over 27,000 Kagu. However, exluding the 5 km of sink habitat around settlements due to dog predation and applying a saturation of the population density at 23% ± 9%, which we observed in protected areas not managed for dogs, a realistic estimate of current Kagu numbers would be 3640 ± 1345. Only 12% of predicted Kagu habitat falls within protected areas, and only half of this protected area is under dog control. Moreover, only 30% of protected rainforest is situated on non-ultramafic soils, which are most suitable for Kagu and can support the highest densities. Numbers of Kagu and other endangered rainforest species could be significantly increased by establishing new protected areas combined with appropriate dog control. We recommend creating 250 km 2 of new protected areas in rainforest on non-ultramafic soils, which could potentially boost the Kagu population by 9000 individuals.
- 20 August 2025
- Published by Springer Nature in Journal of Ornithology
- Vol. 167 (1) , 267-279
Open AccessDensity distribution of the charismatic Kagu to guide conservation of endangered endemic rainforest species in New Caledonia
The Kagu ( Rhynochetos jubatus ) is an endangered endemic bird and emblem of New Caledonia, thus an ideal flagship species for conservation management and planning of protected areas on Grande Terre, the main island. We assessed the density distribution of Kagu by combining results of analyses at two spatial scales. At the local scale, we estimated density in two protected Kagu strongholds. At the island-wide scale, we modelled distribution using MaxEnt over the entire Kagu range on Grande Terre. We then combined these approaches to convert distribution probability into a density distribution. Proportion of rainforest and size of the largest rainforest patch within 250 m × 250 m raster cells were the main factors predicting Kagu distribution. Another important factor was distance to settlements, which is likely related to dog presence, the only significant predator of Kagu. The model identified nearly 2000 km 2 as habitat, which could potentially support over 27,000 Kagu. However, exluding the 5 km of sink habitat around settlements due to dog predation and applying a saturation of the population density at 23% ± 9%, which we observed in protected areas not managed for dogs, a realistic estimate of current Kagu numbers would be 3640 ± 1345. Only 12% of predicted Kagu habitat falls within protected areas, and only half of this protected area is under dog control. Moreover, only 30% of protected rainforest is situated on non-ultramafic soils, which are most suitable for Kagu and can support the highest densities. Numbers of Kagu and other endangered rainforest species could be significantly increased by establishing new protected areas combined with appropriate dog control. We recommend creating 250 km 2 of new protected areas in rainforest on non-ultramafic soils, which could potentially boost the Kagu population by 9000 individuals.
- Open Access
- 13 February 2025
- Published by Elsevier in Animal Behaviour
- Vol. 221
Begging call mimicry and formation of host-specific lineages in the shining bronze-cuckoo, Chalcites lucidus
- 13 February 2025
- Published by Elsevier in Animal Behaviour
- Vol. 221
Open AccessBegging call mimicry and formation of host-specific lineages in the shining bronze-cuckoo, Chalcites lucidus
- Open Access
Advancing cooperative breeding research with a peer-reviewed and “live” Cooperative-Breeding Database (Co-BreeD)
Research on cooperative breeding (a system with the core characteristic of individuals providing care for the offspring of others) is important for understanding sociality and cooperation. However, large-scale comparative analyses on the drivers and consequences of cooperation frequently use considerably inaccurate datasets (e.g. due to inconsistent definitions and outdated information). To advance comparative research on cooperative breeding, we introduce the Co operative- Bree ding D atabase (Co-BreeD), a growing database of key socio-biological parameters of birds and mammals. First, we describe Co-BreeD’s structure as a (i) sample-based (i.e. multiple samples per species linked to an exact sampling location and period), (ii) peer-reviewed and (iii) updatable resource. Respectively, these curating principles allow for (i) investigating intra- and inter-species variation and linking between fine-scale social and environmental parameters, (ii) accuracy and (iii) continuous correction and expansion with the publication of new data. Second, we present the first Co-BreeD dataset, which estimates the prevalence of breeding events with potential alloparents in 265 samples from 233 populations of 150 species, including 2 human societies (N = 26,366 breeding events). We conclude by demonstrating (i) how Co-BreeD facilitates more accurate comparative research (e.g. increased explanatory power by enabling the study of cooperative breeding as a continuous trait, and statistically accounting for the sampling error probabilities), and (ii) that cooperative breeding in birds and mammals is more prevalent than currently estimated.
Open AccessAdvancing cooperative breeding research with a peer-reviewed and “live” Cooperative-Breeding Database (Co-BreeD)
Research on cooperative breeding (a system with the core characteristic of individuals providing care for the offspring of others) is important for understanding sociality and cooperation. However, large-scale comparative analyses on the drivers and consequences of cooperation frequently use considerably inaccurate datasets (e.g. due to inconsistent definitions and outdated information). To advance comparative research on cooperative breeding, we introduce the Co operative- Bree ding D atabase (Co-BreeD), a growing database of key socio-biological parameters of birds and mammals. First, we describe Co-BreeD’s structure as a (i) sample-based (i.e. multiple samples per species linked to an exact sampling location and period), (ii) peer-reviewed and (iii) updatable resource. Respectively, these curating principles allow for (i) investigating intra- and inter-species variation and linking between fine-scale social and environmental parameters, (ii) accuracy and (iii) continuous correction and expansion with the publication of new data. Second, we present the first Co-BreeD dataset, which estimates the prevalence of breeding events with potential alloparents in 265 samples from 233 populations of 150 species, including 2 human societies (N = 26,366 breeding events). We conclude by demonstrating (i) how Co-BreeD facilitates more accurate comparative research (e.g. increased explanatory power by enabling the study of cooperative breeding as a continuous trait, and statistically accounting for the sampling error probabilities), and (ii) that cooperative breeding in birds and mammals is more prevalent than currently estimated.
- Open Access
- 18 July 2023
- Published by Springer Nature in International Journal of Computer Vision
- Vol. 131 (11) , 3118
Correction: Towards Automated Ethogramming: Cognitively-Inspired Event Segmentation for Streaming Wildlife Video Monitoring
- 18 July 2023
- Published by Springer Nature in International Journal of Computer Vision
- Vol. 131 (11) , 3118
Open AccessCorrection: Towards Automated Ethogramming: Cognitively-Inspired Event Segmentation for Streaming Wildlife Video Monitoring
- Open Access
- 28 April 2023
- Published by Springer Nature in International Journal of Computer Vision
- Vol. 131 (9) , 2267-2297
Towards Automated Ethogramming: Cognitively-Inspired Event Segmentation for Streaming Wildlife Video Monitoring
Advances in visual perceptual tasks have been mainly driven by the amount, and types, of annotations of large-scale datasets. Researchers have focused on fully-supervised settings to train models using offline epoch-based schemes. Despite the evident advancements, limitations and cost of manually annotated datasets have hindered further development for event perceptual tasks, such as detection and localization of objects and events in videos. The problem is more apparent in zoological applications due to the scarcity of annotations and length of videos-most videos are at most ten minutes long. Inspired by cognitive theories, we present a self-supervised perceptual prediction framework to tackle the problem of temporal event segmentation by building a stable representation of event-related objects. The approach is simple but effective. We rely on LSTM predictions of high-level features computed by a standard deep learning backbone. For spatial segmentation, the stable representation of the object is used by an attention mechanism to filter the input features before the prediction step. The self-learned attention maps effectively localize the object as a side effect of perceptual prediction. We demonstrate our approach on long videos from continuous wildlife video monitoring, spanning multiple days at 25 FPS. We aim to facilitate automated ethogramming by detecting and localizing events without the need for labels. Our approach is trained in an online manner on streaming input and requires only a single pass through the video, with no separate training set. Given the lack of long and realistic (includes real-world challenges) datasets, we introduce a new wildlife video dataset–nest monitoring of the Kagu (a flightless bird from New Caledonia)–to benchmark our approach. Our dataset features a video from 10 days (over 23 million frames) of continuous monitoring of the Kagu in its natural habitat. We annotate every frame with bounding boxes and event labels. Additionally, each frame is annotated with time-of-day and illumination conditions. We will make the dataset, which is the first of its kind, and the code available to the research community. We find that the approach significantly outperforms other self-supervised, traditional (e.g., Optical Flow, Background Subtraction) and NN-based (e.g., PA-DPC, DINO, iBOT), baselines and performs on par with supervised boundary detection approaches (i.e., PC). At a recall rate of 80%, our best performing model detects one false positive activity every 50 min of training. On average, we at least double the performance of self-supervised approaches for spatial segmentation. Additionally, we show that our approach is robust to various environmental conditions (e.g., moving shadows). We also benchmark the framework on other datasets (i.e., Kinetics-GEBD, TAPOS) from different domains to demonstrate its generalizability. The data and code are available on our project page: https://aix.eng.usf.edu/research_automated_ethogramming.html
- 28 April 2023
- Published by Springer Nature in International Journal of Computer Vision
- Vol. 131 (9) , 2267-2297
Open AccessTowards Automated Ethogramming: Cognitively-Inspired Event Segmentation for Streaming Wildlife Video Monitoring
Advances in visual perceptual tasks have been mainly driven by the amount, and types, of annotations of large-scale datasets. Researchers have focused on fully-supervised settings to train models using offline epoch-based schemes. Despite the evident advancements, limitations and cost of manually annotated datasets have hindered further development for event perceptual tasks, such as detection and localization of objects and events in videos. The problem is more apparent in zoological applications due to the scarcity of annotations and length of videos-most videos are at most ten minutes long. Inspired by cognitive theories, we present a self-supervised perceptual prediction framework to tackle the problem of temporal event segmentation by building a stable representation of event-related objects. The approach is simple but effective. We rely on LSTM predictions of high-level features computed by a standard deep learning backbone. For spatial segmentation, the stable representation of the object is used by an attention mechanism to filter the input features before the prediction step. The self-learned attention maps effectively localize the object as a side effect of perceptual prediction. We demonstrate our approach on long videos from continuous wildlife video monitoring, spanning multiple days at 25 FPS. We aim to facilitate automated ethogramming by detecting and localizing events without the need for labels. Our approach is trained in an online manner on streaming input and requires only a single pass through the video, with no separate training set. Given the lack of long and realistic (includes real-world challenges) datasets, we introduce a new wildlife video dataset–nest monitoring of the Kagu (a flightless bird from New Caledonia)–to benchmark our approach. Our dataset features a video from 10 days (over 23 million frames) of continuous monitoring of the Kagu in its natural habitat. We annotate every frame with bounding boxes and event labels. Additionally, each frame is annotated with time-of-day and illumination conditions. We will make the dataset, which is the first of its kind, and the code available to the research community. We find that the approach significantly outperforms other self-supervised, traditional (e.g., Optical Flow, Background Subtraction) and NN-based (e.g., PA-DPC, DINO, iBOT), baselines and performs on par with supervised boundary detection approaches (i.e., PC). At a recall rate of 80%, our best performing model detects one false positive activity every 50 min of training. On average, we at least double the performance of self-supervised approaches for spatial segmentation. Additionally, we show that our approach is robust to various environmental conditions (e.g., moving shadows). We also benchmark the framework on other datasets (i.e., Kinetics-GEBD, TAPOS) from different domains to demonstrate its generalizability. The data and code are available on our project page: https://aix.eng.usf.edu/research_automated_ethogramming.html
- 11 December 2022
- Published by Wiley in Journal of Animal Ecology
- Vol. 92 (1) , 30-43
Polymorphism at the nestling stage and host‐specific mimicry in an Australasian cuckoo‐host arms race
Decades of research have shown that the coevolutionary arms race between avian brood parasites and their hosts can promote phenotypic diversification in hosts and brood parasites. However, relatively little is known about the role of brood parasitism in promoting phenotypic diversification of nestlings. We review field data collected over four decades in Australia, New Caledonia and New Zealand to assess potential for coevolutionary interactions between the shining bronze‐cuckoo (Chalcites lucidus) and its hosts, and how diversification at the nestling stage may be generating different subspecies. The shining bronze‐cuckoo is a specialist parasite of a few hosts in the family Acanthizidae. It has diversified into subspecies, of which the nestlings closely mimic the respective host nestlings in each region. Additionally, some cuckoo subspecies have polymorphic nestlings. The Acanthizidae hosts have similar breeding and nesting habits and only moderately effective frontline defences against parasitism at cuckoo egg laying or at the egg stages. However, some hosts have developed highly effective defences at the nestling stage by recognising and ejecting cuckoo nestlings from the nest. As with the cuckoo nestlings, some hosts have polymorphic nestlings. The coevolutionary interactions in each region suggest different evolutionary stages of the arms race in which either the parasite or the host is currently in the lead. The presence of moderately effective defences at the egg laying and egg stages might explain why some hosts do not have defences at the nestling stage. The south‐Pacific cuckoo – host systems are excellent models to explore the evolutionary mechanisms driving the diversification at the nestling stage in the coevolutionary arms race between avian brood parasites and their hosts.
- 11 December 2022
- Published by Wiley in Journal of Animal Ecology
- Vol. 92 (1) , 30-43
Polymorphism at the nestling stage and host‐specific mimicry in an Australasian cuckoo‐host arms race
Decades of research have shown that the coevolutionary arms race between avian brood parasites and their hosts can promote phenotypic diversification in hosts and brood parasites. However, relatively little is known about the role of brood parasitism in promoting phenotypic diversification of nestlings. We review field data collected over four decades in Australia, New Caledonia and New Zealand to assess potential for coevolutionary interactions between the shining bronze‐cuckoo (Chalcites lucidus) and its hosts, and how diversification at the nestling stage may be generating different subspecies. The shining bronze‐cuckoo is a specialist parasite of a few hosts in the family Acanthizidae. It has diversified into subspecies, of which the nestlings closely mimic the respective host nestlings in each region. Additionally, some cuckoo subspecies have polymorphic nestlings. The Acanthizidae hosts have similar breeding and nesting habits and only moderately effective frontline defences against parasitism at cuckoo egg laying or at the egg stages. However, some hosts have developed highly effective defences at the nestling stage by recognising and ejecting cuckoo nestlings from the nest. As with the cuckoo nestlings, some hosts have polymorphic nestlings. The coevolutionary interactions in each region suggest different evolutionary stages of the arms race in which either the parasite or the host is currently in the lead. The presence of moderately effective defences at the egg laying and egg stages might explain why some hosts do not have defences at the nestling stage. The south‐Pacific cuckoo – host systems are excellent models to explore the evolutionary mechanisms driving the diversification at the nestling stage in the coevolutionary arms race between avian brood parasites and their hosts.
- Open Access
- 29 July 2022
- Published by Springer Nature in Biological Invasions
- Vol. 24 (12) , 3675-3680
Impact of invasive little fire ants Wasmannia auropunctata on rainforest soil fauna: implications for conservation of the endangered flightless kagu of New Caledonia
The little fire ant, Wasmannia auropunctata, is an invasive species known to negatively affect many native species in its introduced range. Here, we studied the effect of little fire ant presence on the biomass of soil fauna in rainforest of New Caledonia to assess if little fire ants have an impact on an endangered endemic flightless bird, the kagu, Rhynochetos jubatus. The average biomass of soil fauna (except for millipedes) was lower in 53 sample plots with little fire ants than in 597 sample plots without little fire ants and the biomass decreased with increasing density of little fire ant nests. Millipedes on the contrary seemed to be positively impacted as their average biomass was higher in areas with fire ants than in areas without, but even millipede biomass decreased with increasing nest density of fire ants. Kagu density decreased with increasing area of rainforest invaded by little fire ants, suggesting, given their observed negative impact on soil fauna biomass, that they may limit food resources for kagu. Little fire ants might therefore transform rainforest into a less suitable habitat for the kagu by reducing the biomass of their prey and thus have a negative impact for kagu conservation.
- 29 July 2022
- Published by Springer Nature in Biological Invasions
- Vol. 24 (12) , 3675-3680
Open AccessImpact of invasive little fire ants Wasmannia auropunctata on rainforest soil fauna: implications for conservation of the endangered flightless kagu of New Caledonia
The little fire ant, Wasmannia auropunctata, is an invasive species known to negatively affect many native species in its introduced range. Here, we studied the effect of little fire ant presence on the biomass of soil fauna in rainforest of New Caledonia to assess if little fire ants have an impact on an endangered endemic flightless bird, the kagu, Rhynochetos jubatus. The average biomass of soil fauna (except for millipedes) was lower in 53 sample plots with little fire ants than in 597 sample plots without little fire ants and the biomass decreased with increasing density of little fire ant nests. Millipedes on the contrary seemed to be positively impacted as their average biomass was higher in areas with fire ants than in areas without, but even millipede biomass decreased with increasing nest density of fire ants. Kagu density decreased with increasing area of rainforest invaded by little fire ants, suggesting, given their observed negative impact on soil fauna biomass, that they may limit food resources for kagu. Little fire ants might therefore transform rainforest into a less suitable habitat for the kagu by reducing the biomass of their prey and thus have a negative impact for kagu conservation.
- 29 July 2022
- Published by CSIRO Publishing in Australian Mammalogy
- Vol. 45 (1) , 121-124
A trial of satellite GPS telemetry on feral pigs in tropical mountain rainforest
We deployed two GPS collars and one innovative harness-mounted GPS backpack with Iridium or Globalstar satellite data transmission on feral pigs in mountain rainforest with deep valleys in New Caledonia during the hot wet season. Both collar and backpack with Iridium transmission achieved high success rates of locations and the rate increased with decreasing interval length between locations. The collar with Globalstar transmission had very low fix rates at 13-h intervals and was insufficient to produce robust results. The backpack harness was still functional after 20 months of use and proved to be a reliable alternative to collars.
- 29 July 2022
- Published by CSIRO Publishing in Australian Mammalogy
- Vol. 45 (1) , 121-124
A trial of satellite GPS telemetry on feral pigs in tropical mountain rainforest
We deployed two GPS collars and one innovative harness-mounted GPS backpack with Iridium or Globalstar satellite data transmission on feral pigs in mountain rainforest with deep valleys in New Caledonia during the hot wet season. Both collar and backpack with Iridium transmission achieved high success rates of locations and the rate increased with decreasing interval length between locations. The collar with Globalstar transmission had very low fix rates at 13-h intervals and was insufficient to produce robust results. The backpack harness was still functional after 20 months of use and proved to be a reliable alternative to collars.