Learning from algorithm-generated pseudo-annotations for detecting ants in videos.

Deep learning (DL) based detection models are powerful tools for large-scale analysis of dynamic biological behaviors in video data. Supervised training of a DL detection model often requires a large amount of manually-labeled training data which are time-consuming and labor-intensive to acquire. In this paper, we propose LFAGPA (Learn From Algorithm-Generated Pseudo-Annotations) that utilizes (noisy) annotations which are automatically generated by algorithms to train DL models for ant detection in videos. Our method consists of two main steps: (1) generate foreground objects using a (set of) state-of-the-art foreground extraction algorithm(s); (2) treat the results from step (1) as pseudo-annotations and use them to train deep neural networks for ant detection. We tackle several challenges on how to make use of automatically generated noisy annotations, how to learn from multiple annotation resources, and how to combine algorithm-generated annotations with human-labeled annotations (when available) for this learning framework. In experiments, we evaluate our method using 82 videos (totally 20,348 image frames) captured under natural conditions in a tropical rain-forest for dynamic ant behavior study. Without any manual annotation cost but only algorithm-generated annotations, our method can achieve a decent detection performance (77% in [Formula: see text] score). Moreover, when using only 10% manual annotations, our method can train a DL model to perform as well as using the full human annotations (81% in [Formula: see text] score).

Bumble bee colony health and performance vary widely across the urban ecosystem.

Urbanization is a global phenomenon that can affect fitness and could challenge the persistence of most species, including wild bee pollinators. Yet, how and which environmental features affect bee health and fitness within the urban ecosystem remain unclear. Here, we placed experimental Bombus terrestris colonies in sites spanning from the edge into a city's core to investigate bumble bee parasitism, foraging behaviour, energetic stress, colony growth and reproductive output. In each site, ambient temperature was recorded, the availability of floral resources was evaluated and landscape heterogeneity was characterized using land-cover maps. We found that Bombus terrestris parasitism levels increased across the season in line with colony growth but were negatively related to the proportion of impervious surfaces surrounding a site. Bombus terrestris foraging trip duration decreased with increasing ecotones (edge density) but, conversely, increased in sites with honey bee hives present. Energetic stress was evaluated as lowered trehalose titre in the haemolymph of returning foragers; stress increased with the proportion of impervious surfaces. Furthermore, our analyses identified ambient temperature to be a strong predictor of Bombus terrestris colony performance in that high ambient temperature reduced colony growth and indirectly the production of sexual offspring (gynes). Our results highlight the importance of ecotones as well as minimizing the intensity of urbanization and urban honey bee beekeeping for bumble bee colony health and foraging behaviour. They also point to the importance of microclimate (i.e. temperature) for bumble bee colony performance and suggest that increasing temperatures could have a negative impact in slowing colony weight gain, and indirectly in reducing colony reproduction.

Evaluation of direct and indirect transmission of fungal spores in ants.

Parasite transmission plays a central role in disease dynamics, but little is known about the extent to which direct and indirect transmission contributes to disease dynamics in group-living animals. Quantifying transmission by contact or exposure is challenging, as direct measurements of pathogen transmission are often impractical and individual behaviour is largely unknown. Here, we attempt to tackle these twin problems through the use of a generalist entomopathogenic fungus Metarhizium robertsii to study direct and indirect transmission in the black carpenter ant Camponotus pennsylvanicus. We provide evidence that large quantities of fungal conidia (spores) can be acquired over time on the ant's cuticle and significant amounts are shed into the environment, allowing indirect transmission. Unexposed ants in the nest can be infected either through direct (i.e. social contacts) or indirect (i.e. acquisition from the environment) transmission, potentially leading to the same mortality as if all individuals of a nest would have been directly exposed to the pathogen. Our findings highlight the importance of indirect transmission routes in a species of social living organisms.

Automated tracking and analysis of ant trajectories shows variation in forager exploration.

Determining how ant colonies optimize foraging while mitigating pathogen and predator risks provides insight into how the ants have achieved ecological success. Ants must respond to changing resource conditions, but exploration comes at a cost of higher potential exposure to threats. Fungal infected cadavers surround the main foraging trails of the carpenter ant Camponotus rufipes, offering a system to study how foragers behave given the persistent occurrence of disease threats. Studies on social insect foraging behavior typically require many hours of human labor due to the high density of individuals. To overcome this, we developed deep learning based computer vision algorithms to track foraging ants, frame-by-frame, from video footage shot under the natural conditions of a tropical forest floor at night. We found that most foragers walk in straight lines overlapping the same areas as other ants, but there is a subset of foragers with greater exploration. Consistency in walking behavior may protect most ants from infection, while foragers that explore unique portions of the trail may be more likely to encounter fungal spores implying a trade-off between resource discovery and risk avoidance.

Infection dynamics of Nosema ceranae in honey bee midgut and host cell apoptosis.

Nosema ceranae is an intracellular microsporidian parasite that infects epithelial cells of the honey bee (Apis mellifera) midgut. Previous studies have shown that Nosema may alter cell renewal and apoptosis in honey bees. We found that the amount of apoptotic cells progressively declines from the anterior towards posterior regions of the midgut in Nosema-infected sensitive bees. There was no such pattern in the infected Nosema tolerant honey bees and controls. These data provide additional evidence that N. ceranae appears to alter apoptosis in its host cells for its own advantage.

"Up" or "down" that makes the difference. How giant honeybees (Apis dorsata) see the world.

A. dorsata builds its large exposed comb high in trees or under ledges of high rocks. The "open" nest of A. dorsata, shielded (only!) by multiple layers of bees, is highly vulnerable to any kind of direct contact or close range attacks from predators. Therefore, guard bees of the outer layer of A. dorsata's nest monitor the vicinity for possible hazards and an effective risk assessment is required. Guard bees, however, are frequently exposed to different objects like leaves, twigs and other tree litter passing the nest from above and falling to the ground. Thus, downward movement of objects past the nest might be used by A. dorsata to classify these visual stimuli near the nest as "harmless". To test the effect of movement direction on defensive responses, we used circular black discs that were moved down or up in front of colonies and recorded the number of guard bees flying towards the disc. The size of the disc (diameter from 8 cm to 50 cm) had an effect on the number of guard bees responding, the bigger the plate the more bees started from the nest. The direction of a disc's movement had a dramatic effect on the attraction. We found a significantly higher number of attacks, when discs were moved upwards compared to downward movements (GLMM (estimate ± s.e.) 1.872 ± 0.149, P < 0.001). Our results demonstrate for the first time that the vertical direction of movement of an object can be important for releasing defensive behaviour. Upward movement of dark objects near the colony might be an innate releaser of attack flights. At the same time, downward movement is perceived as a "harmless" stimulus.

Differential proteomics reveals novel insights into Nosema-honey bee interactions.

Host manipulation is a common strategy by parasites to reduce host defense responses, enhance development, host exploitation, reproduction and, ultimately, transmission success. As these parasitic modifications can reduce host fitness, increased selection pressure may result in reciprocal adaptations of the host. Whereas the majority of studies on host manipulation have explored resistance against parasites (i.e. ability to prevent or limit an infection), data describing tolerance mechanisms (i.e. ability to limit harm of an infection) are scarce. By comparing differential protein abundance, we provide evidence of host-parasite interactions in the midgut proteomes of N. ceranae-infected and uninfected honey bees from both Nosema-tolerant and Nosema-sensitive lineages. We identified 16 proteins out of 661 protein spots that were differentially abundant between experimental groups. In general, infections of Nosema resulted in an up-regulation of the bee's energy metabolism. Additionally, we identified 8 proteins that were differentially abundant between tolerant and sensitive honey bees regardless of the Nosema infection. Those proteins were linked to metabolism, response to oxidative stress and apoptosis. In addition to bee proteins, we also identified 3 Nosema ceranae proteins. Interestingly, abundance of two of these Nosema proteins were significantly higher in infected Nosema-sensitive honeybees relative to the infected Nosema-tolerant lineage. This may provide a novel candidate for studying the molecular interplay between N. ceranae and its honey bee host in more detail.

Parasite resistance and tolerance in honeybees at the individual and social level.

Organisms living in large groups, such as social insects, are particularly vulnerable to parasite transmission. However, they have evolved diverse defence mechanisms which are not only restricted to the individual's immune response, but also include social defences. Here, we review cases of adaptations at the individual and social level in the honeybee Apis mellifera against the ectoparasitic mite Varroa destructor and the endoparasitic microsporidians Nosema ceranae and Nosema apis. They are considered important threats to honeybee health worldwide. We highlight how individual resistance may result in tolerance at the colony level and vice versa.

Nosema spp. infections cause no energetic stress in tolerant honeybees.

Host-pathogen coevolution leads to reciprocal adaptations, allowing pathogens to increase host exploitation or hosts to minimise costs of infection. As pathogen resistance is often associated with considerable costs, tolerance may be an evolutionary alternative. Here, we examined the effect of two closely related and highly host dependent intracellular gut pathogens, Nosema apis and Nosema ceranae, on the energetic state in Nosema tolerant and sensitive honeybees facing the infection. We quantified the three major haemolymph carbohydrates fructose, glucose, and trehalose using high-performance liquid chromatography (HPLC) as a measure for host energetic state. Trehalose levels in the haemolymph were negatively associated with N. apis infection intensity and with N. ceranae infection regardless of the infection intensity in sensitive honeybees. Nevertheless, there was no such association in Nosema spp. infected tolerant honeybees. These findings suggest that energy availability in tolerant honeybees was not compromised by the infection. This result obtained at the individual level may also have implications at the colony level where workers in spite of a Nosema infection can still perform as well as healthy bees, maintaining colony efficiency and productivity.

Nosema Tolerant Honeybees (Apis mellifera) Escape Parasitic Manipulation of Apoptosis.

Apoptosis is not only pivotal for development, but also for pathogen defence in multicellular organisms. Although numerous intracellular pathogens are known to interfere with the host's apoptotic machinery to overcome this defence, its importance for host-parasite coevolution has been neglected. We conducted three inoculation experiments to investigate in the apoptotic respond during infection with the intracellular gut pathogen Nosema ceranae, which is considered as potential global threat to the honeybee (Apis mellifera) and other bee pollinators, in sensitive and tolerant honeybees. To explore apoptotic processes in the gut epithelium, we visualised apoptotic cells using TUNEL assays and measured the relative expression levels of subset of candidate genes involved in the apoptotic machinery using qPCR. Our results suggest that N. ceranae reduces apoptosis in sensitive honeybees by enhancing inhibitor of apoptosis protein-(iap)-2 gene transcription. Interestingly, this seems not be the case in Nosema tolerant honeybees. We propose that these tolerant honeybees are able to escape the manipulation of apoptosis by N. ceranae, which may have evolved a mechanism to regulate an anti-apoptotic gene as key adaptation for improved host invasion.

Factors influencing Dipylidium sp. infection in a free-ranging social carnivore, the spotted hyaena (Crocuta crocuta).

We provide the first genetic sequence data for a Dipylidium species from a wild carnivore plus an analysis of the effects of ecological, demographic, physiological and behavioural factors on Dipylidium sp. infection prevalence in a social carnivore, the spotted hyaena (Crocuta crocuta), in the Serengeti National Park, Tanzania. Our sequence data from a mitochondrial gene fragment (1176 base pair long) had a similarity of between 99% and 89% to Dipylidium caninum. We determined infection prevalence in 146 faecal samples from 124 known animals in three social groups (termed clans) using molecular screening and Dipylidium proglottid presence. Our analysis revealed significantly higher infection prevalence in juveniles (55%) than adults (15.8%), indicating that predominantly juveniles maintained infection in clans. The likelihood of infection in juveniles significantly: (1) increased as the number of adults and older juveniles (>6 months) at communal dens increased, implying a positive relationship between this factor and the size of the intermediate host (probably a flea species) population at communal dens; (2) decreased as the number of younger juveniles (<6 months) increased, suggesting that the chance of susceptible juveniles ingesting infected fleas during self-grooming declined as the number of infected fleas per younger juvenile declined; and (3) decreased during periods of low prey abundance in clan territories when an increased reliance on long-distances foraging excursions reduces the number of clan members visiting communal dens, possibly resulting in a decline in flea populations at dens. Long-distance foraging also increases the intervals (in days) between nursing visits by lactating females to their offspring. Lengthy intervals between milk intake by infected juveniles may reduce adult Dipylidium fecundity and hence decrease infection prevalence in the den flea population. Our study provides useful insights into Dipylidium epidemiology in a social carnivore population subject to large fluctuations in prey abundance.