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).

Portable Diffuse Reflectance Spectroscopy of Potato Leaves for Pre-Symptomatic Detection of Late Blight Disease.

We report on the use of leaf diffuse reflectance spectroscopy for plant disease detection. A smartphone-operated, compact diffused reflectance spectrophotometer is used for field collection of leaf diffuse reflectance spectra to enable pre-symptomatic detection of the progression of potato late blight disease post inoculation with oomycete pathogen Phytophthora infestans. Neural-network-based analysis predicts infection with >96% accuracy, only 24 h after inoculation with the pathogen, and nine days before visual late blight symptoms appear. Our study demonstrates the potential of using portable optical spectroscopy in tandem with machine learning analysis for early diagnosis of plant diseases.

Potential distribution of fall armyworm in Africa and beyond, considering climate change and irrigation patterns.

The fall armyworm, Spodoptera frugiperda (FAW), first invaded Africa in 2016 and has since become established in many areas across the continent where it poses a serious threat to food and nutrition security. We re-parameterized the existing CLIMEX model to assess the FAW global invasion threat, emphasizing the risk of transient and permanent population establishment in Africa under current and projected future climates, considering irrigation patterns. FAW can establish itself in almost all countries in eastern and central Africa and a large part of western Africa under the current climate. Climatic barriers, such as heat and dry stresses, may limit the spread of FAW to North and South Africa. Future projections suggest that FAW invasive range will retract from both northern and southern regions towards the equator. However, a large area in eastern and central Africa is projected to have an optimal climate for FAW persistence. These areas will serve as FAW 'hotspots' from where it may migrate to the north and south during favorable seasons and then pose an economic threat. Our projections can be used to identify countries at risk for permanent and transient FAW-population establishment and inform timely integrated pest management interventions under present and future climate in Africa.

Social fluidity mobilizes contagion in human and animal populations.

Humans and other group-living animals tend to distribute their social effort disproportionately. Individuals predominantly interact with a small number of close companions while maintaining weaker social bonds with less familiar group members. By incorporating this behavior into a mathematical model, we find that a single parameter, which we refer to as social fluidity, controls the rate of social mixing within the group. Large values of social fluidity correspond to gregarious behavior, whereas small values signify the existence of persistent bonds between individuals. We compare the social fluidity of 13 species by applying the model to empirical human and animal social interaction data. To investigate how social behavior influences the likelihood of an epidemic outbreak, we derive an analytical expression of the relationship between social fluidity and the basic reproductive number of an infectious disease. For species that form more stable social bonds, the model describes frequency-dependent transmission that is sensitive to changes in social fluidity. As social fluidity increases, animal-disease systems become increasingly density-dependent. Finally, we demonstrate that social fluidity is a stronger predictor of disease outcomes than both group size and connectivity, and it provides an integrated framework for both density-dependent and frequency-dependent transmission.

An agent-based model shows zombie ants exhibit search behavior.

Parasites can alter the behavior of animals. Such alterations could be a byproduct of infection or actively controlled and directed by the parasite. Ants infected with zombie ant fungi (Ophiocordyceps sp.) show behavioral changes culminating in the ant dying while biting into vegetation. To investigate the influence of the parasite on behavioral changes, we created an agent-based model that provides a prediction of how fungal infected ants move before death. The model shows how alterations in movement, such as an increased turning rate, within the normal range of ant behavior, can lead a host from the nest to the underside of a leaf. This demonstrates the simplicity in how such behavioral changes could evolve, as the fungal parasite could benefit from the natural behavior of the host, contesting a hypothesis of highly directed manipulation.

Insect Behavioral Change and the Potential Contributions of Neuroinflammation-A Call for Future Research.

Many organisms are able to elicit behavioral change in other organisms. Examples include different microbes (e.g., viruses and fungi), parasites (e.g., hairworms and trematodes), and parasitoid wasps. In most cases, the mechanisms underlying host behavioral change remain relatively unclear. There is a growing body of literature linking alterations in immune signaling with neuron health, communication, and function; however, there is a paucity of data detailing the effects of altered neuroimmune signaling on insect neuron function and how glial cells may contribute toward neuron dysregulation. It is important to consider the potential impacts of altered neuroimmune communication on host behavior and reflect on its potential role as an important tool in the "neuro-engineer" toolkit. In this review, we examine what is known about the relationships between the insect immune and nervous systems. We highlight organisms that are able to influence insect behavior and discuss possible mechanisms of behavioral manipulation, including potentially dysregulated neuroimmune communication. We close by identifying opportunities for integrating research in insect innate immunity, glial cell physiology, and neurobiology in the investigation of behavioral manipulation.

A new zombie ant behavior unraveled: Aggregating on tree trunks.

Hosts can be manipulated by parasites to move to locations advantageous for onward transmission. To investigate the role of behavioral manipulation in creating transmission hotspots, we studied the distribution of zombie turtle ants in the Amazon rainforest. The turtle ant Cephalotes atratus nests and mostly forages in the canopy, but is found at the base of trees when infected with the zombie ant fungus Ophiocordyceps kniphofioides. We found 626 infected cadavers on 14.8% of 162 trees sampled. Cadavers were highly aggregated on the surface of the trees, explained by behavioral observations indicating infected ants as slightly attracted to zombie ant cadavers on a tree. From 1,726 h of camera footage, we recorded the removal of three zombie ant cadavers by live ants. The number of removals compared to the density of infected individuals indicates the base of a tree as an escape from the evolved ability of social insects to recognize and treat disease inside the nest, allowing the parasite to continuously remain in the environment.

Accuracy of a Smartphone-Based Object Detection Model, PlantVillage Nuru, in Identifying the Foliar Symptoms of the Viral Diseases of Cassava-CMD and CBSD.

Nuru is a deep learning object detection model for diagnosing plant diseases and pests developed as a public good by PlantVillage (Penn State University), FAO, IITA, CIMMYT, and others. It provides a simple, inexpensive and robust means of conducting in-field diagnosis without requiring an internet connection. Diagnostic tools that do not require the internet are critical for rural settings, especially in Africa where internet penetration is very low. An investigation was conducted in East Africa to evaluate the effectiveness of Nuru as a diagnostic tool by comparing the ability of Nuru, cassava experts (researchers trained on cassava pests and diseases), agricultural extension officers and farmers to correctly identify symptoms of cassava mosaic disease (CMD), cassava brown streak disease (CBSD) and the damage caused by cassava green mites (CGM). The diagnosis capability of Nuru and that of the assessed individuals was determined by inspecting cassava plants and by using the cassava symptom recognition assessment tool (CaSRAT) to score images of cassava leaves, based on the symptoms present. Nuru could diagnose symptoms of cassava diseases at a higher accuracy (65% in 2020) than the agricultural extension agents (40-58%) and farmers (18-31%). Nuru's accuracy in diagnosing cassava disease and pest symptoms, in the field, was enhanced significantly by increasing the number of leaves assessed to six leaves per plant (74-88%). Two weeks of Nuru practical use provided a slight increase in the diagnostic skill of extension workers, suggesting that a longer duration of field experience with Nuru might result in significant improvements. Overall, these findings suggest that Nuru can be an effective tool for in-field diagnosis of cassava diseases and has the potential to be a quick and cost-effective means of disseminating knowledge from researchers to agricultural extension agents and farmers, particularly on the identification of disease symptoms and their management practices.

Zombie-Ant Fungi Emerged from Non-manipulating, Beetle-Infecting Ancestors.

The manipulation of animal behavior by parasitic organisms is one of the most complex adaptations to have arisen via natural selection. Among the most impressive examples of behavioral manipulation are the zombie-ant fungi [1]. In this association, ants are controlled to leave the colony and perform a stereotyped death grip behavior, where they bite onto vegetation over foraging trails, before being killed for the post mortem fungal growth. Manipulation functions to provide a platform outside the nest, from which fungal parasites actively shoot out spores, targeting foraging ants because within colony transmission is prevented by strong social immunity exhibited by social insect societies [2, 3]. It is not clear how such complex examples of host manipulation arose. To address this, we performed a broad-scale phylogenetic reconstruction of the order Hypocreales, to which the zombie-ant fungi, Ophiocordyceps, belong. In order to understand the patterns of host association and host switching along the evolution of Ophiocordyceps, we performed ancestral character state reconstruction analysis. We found that zombie-ant fungi likely arose from an ancestor that infected beetle larvae residing in soil or decaying wood, similar to extant beetle-infecting Ophiocordyceps species. Surprisingly, the jump led to an extensive species radiation observed after the development of behavioral manipulation. We suggest that the jump from solitary beetle larva to ants within a colony exposed the fungus to the robust social immunity of ant societies.

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.

The metabolic alteration and apparent preservation of the zombie ant brain.

Some parasites can manipulate the behavior of their animal hosts to increase transmission. An interesting area of research is understanding how host neurobiology is manipulated by microbes to the point of displaying such aberrant behaviors. Here, we characterize the metabolic profile of the brain of an insect at the moment of the behavioral manipulation by a parasitic microbe. Our model system are ants infected with the parasitic fungus Ophiocordyceps kimflemingiae (=unilateralis), which manipulates ants to climb and bite into plant substrates, before killing the host (i.e. zombie ants). At the moment of the behavioral manipulation by the fungus, the host's brain is not invaded by the fungus which is known to extensively invade muscle tissue. We found that, despite not being invaded by the parasite, the brains of manipulated ants are notably different, showing alterations in neuromodulatory substances, signs of neurodegeneration, changes in energy use, and antioxidant compound that signal stress reactions by the host. Ergothionine, a fungal derived compound with known neuronal cytoprotection functions was found to be highly elevated in zombie ant brains suggesting the fungus, which does not invade the central nervous system, is preserving the brain.

Zombie ant death grip due to hypercontracted mandibular muscles.

There are numerous examples of parasites that manipulate the behavior of the hosts that they infect. One such host-pathogen relationship occurs between the 'zombie-ant fungus' Ophiocordyceps unilateralis sensu lato and its carpenter ant host. Infected ants climb to elevated locations and bite onto vegetation where they remain permanently affixed well after death. The mandibular muscles, but not the brain, of infected ants are extensively colonized by the fungus. We sought to investigate the mechanisms by which O. unilateralis s.l. may be able to influence mandibular muscle contraction despite widespread muscle damage. We found that infected muscles show evidence of hypercontraction. Despite the extensive colonization, both motor neurons and neuromuscular junctions appear to be maintained. Infection results in sarcolemmal damage, but this is not specific to the death grip. We found evidence of precise penetration of muscles by fungal structures and the presence of extracellular vesicle-like particles, both of which may contribute to mandibular hypercontraction.

Dynamic and game theory of infectious disease stigmas.

Stigmas are a primal phenomena, ubiquitous in human societies past and present. Some evolutionary anthropologists have argued that stigmatization in response to disease is an adaptive behavior because stigmatization may help people and communities reduce the risks they face from infectious diseases and increase reproductive success. On the other hand, some cultural anthropologists and social critics argue that stigmatization has strong negative impacts on community health. One recent analysis resolved this conflict by hypothesizing that stigmas had individual and group-evolutionary benefits in the past but are now maladaptive because of intervening societal transitions. Here, we present a quantitative theory of infectious disease stigmatization. Using a four-compartment model of stigmatization against a chronic disease, we show a stigma ratio, being the ratio of net transmissions by stigmatized people to net transmissions by unstigmatized people, predicts the impact of stigmatization on lifetime infection risk. When stigmatized people are segregated from the rest of the population and there are no alternative interventions that reduce transmission, stigmatization can reduce prevalence and infection risk. When stigmas do not lead to segregation but do discourage behavior change and reduce access to medical interventions, stigmatization acts to increases the lifetime risk of infection in the community. We further show that fear of stigmas can create policy resistance to healthcare access. The societal consequences of fear are worse when effective medical treatment is available. We conclude that stigma's can be adaptive, but good healthcare and leaky ostracism can make stigmas against chronic infectious disease maladaptive, and that the deprecation of stigmas is a natural transition in the modern urban societies.

Ant colonies maintain social homeostasis in the face of decreased density.

Interactions lie at the heart of social organization, particularly in ant societies. Interaction rates are presumed to increase with density, but there is little empirical evidence for this. We manipulated density within carpenter ant colonies of the species Camponotus pennsylvanicus by quadrupling nest space and by manually tracking 6.9 million ant locations and over 3200 interactions to study the relationship between density, spatial organization and interaction rates. Colonies divided into distinct spatial regions on the basis of their underlying spatial organization and changed their movement patterns accordingly. Despite a reduction in both overall and local density, we did not find the expected concomitant reduction in interaction rates across all colonies. Instead, we found divergent effects across colonies. Our results highlight the remarkable organizational resilience of ant colonies to changes in density, which allows them to sustain two key basic colony life functions, that is food and information exchange, during environmental change.

A Mobile-Based Deep Learning Model for Cassava Disease Diagnosis.

Convolutional neural network (CNN) models have the potential to improve plant disease phenotyping where the standard approach is visual diagnostics requiring specialized training. In scenarios where a CNN is deployed on mobile devices, models are presented with new challenges due to lighting and orientation. It is essential for model assessment to be conducted in real world conditions if such models are to be reliably integrated with computer vision products for plant disease phenotyping. We train a CNN object detection model to identify foliar symptoms of diseases in cassava (Manihot esculenta Crantz). We then deploy the model in a mobile app and test its performance on mobile images and video of 720 diseased leaflets in an agricultural field in Tanzania. Within each disease category we test two levels of severity of symptoms-mild and pronounced, to assess the model performance for early detection of symptoms. In both severities we see a decrease in performance for real world images and video as measured with the F-1 score. The F-1 score dropped by 32% for pronounced symptoms in real world images (the closest data to the training data) due to a decrease in model recall. If the potential of mobile CNN models are to be realized our data suggest it is crucial to consider tuning recall in order to achieve the desired performance in real world settings. In addition, the varied performance related to different input data (image or video) is an important consideration for design in real world applications.

Parasite manipulation of host behavior.

Some parasites have evolved the ability to precisely control the behavior of animals in ways that enhance the transmission of parasite genes into the next generation. This is the concept of the 'extended phenotype' first conceived by Richard Dawkins in 1982. It states that the behavior we observe in animals is due not only to the expression of their genes, but also to the genes of parasites infecting them. In such cases, the behavior is an extended phenotype of the parasite.

Evidence for convergent evolution of host parasitic manipulation in response to environmental conditions.

Environmental conditions exert strong selection on animal behavior. We tested the hypothesis that the altered behavior of hosts due to parasitic manipulation is also subject to selection imposed by changes in environmental conditions over time. Our model system is ants manipulated by parasitic fungi to bite onto vegetation. We analyzed the correlation between forest type (tropical vs. temperate) and the substrate where the host bites (biting substrate: leaf vs. twigs), the time required for the fungi to reach reproductive maturity, and the phylogenetic relationship among specimens from tropical and temperate forests from different parts of the globe. We show that fungal development in temperate forests is longer than the period of time leaves are present and the ants are manipulated to bite twigs. When biting twigs, 90% of the dead ants we examined had their legs wrapped around twigs, which appears to provide better attachment to the plant. Ancestral state character reconstruction suggests that leaf biting is the ancestral trait and that twig biting is a convergent trait in temperate regions of the globe. These three lines of evidence suggest that changes in environmental conditions have shaped the manipulative behavior of the host by its parasite.

Within the fortress: A specialized parasite is not discriminated against in a social insect society.

Social insect colonies function cohesively due, in part, to altruistic behaviors performed towards related individuals. These colonies can be affected by parasites in two distinct ways, either at the level of the individual or the entire colony. As such, colonies of social insects can experience conflict with infected individuals reducing the cohesiveness that typifies them. Parasites of social insects therefore offer us a framework to study conflicts within social insect colonies in addition to the traditionally viewed conflicts afforded by groups of low genetic relatedness due to multiple mating for example. In our study, we use the behavior manipulating fungal pathogen, Ophiocordyceps kimflemingiae (= unilateralis) and its host, Camponotus castaneus, to ask if colony members are able to detect infected individuals. Such detection would be optimal for the colony since infected workers die near foraging trails where the fungus develops its external structures and releases spores that infect other colony members. To determine if C. castaneus workers can detect these future threats, we used continuous-time point observations coupled with longer continuous observations to discern any discrimination towards infected individuals. After observing 1,240 hours of video footage we found that infected individuals are not removed from the colony and continuously received food during the course of fungal infection. We also calculated the distances between workers and the nest entrance in a total of 35,691 data points to find infected workers spent more time near the entrance of the nest. Taken together, these results suggest healthy individuals do not detect the parasite inside their nestmates. The colony's inability to detect infected individuals allows O. kimflemingiae to develop within the colony, while receiving food and protection from natural enemies, which could damage or kill its ant host before the parasite has completed its development.

Neuroparasitology of Parasite-Insect Associations.

Insect behavior can be manipulated by parasites, and in many cases, such manipulation involves the central and peripheral nervous system. Neuroparasitology is an emerging branch of biology that deals with parasites that can control the nervous system of their host. The diversity of parasites that can manipulate insect behavior ranges from viruses to macroscopic worms and also includes other insects that have evolved to become parasites (notably, parasitic wasps). It is remarkable that the precise manipulation observed does not require direct entry into the insect brain and can even occur when the parasite is outside the body. We suggest that a spatial view of manipulation provides a holistic approach to examining such interactions. Integration across approaches from natural history to advanced imaging techniques, omics, and experiments will provide new vistas in neuroparasitology. We also suggest that for researchers interested in the proximate mechanisms of insect behaviors, studies of parasites that have evolved to control such behavior is of significant value.