Animals communicate using multiple sensory channels, including via vision. The colourful plumage of birds is a model system to study visual communication, having evolved through a complex interplay of processes, acting not only on the ability of a plumage patch to convey information, but also in response to physiological and environmental factors. Although much research on inter-specific variation in bird plumage has concentrated on sexual selection, much less has considered the role of non-sexual selection and how it is affected by the joint effects of avian viewing conditions and receiver vision. Here, we combined a taxonomically diverse database of avian plumage reflectance measurements with bird vision models, habitat and behavioural data to test the effect of three factors that affect viewing conditions—habitat openness, migratory preference and diel activity—on avian plumage contrast, accounting for shared evolutionary history and variation in avian visual systems. We find that habitat structure and migratory preference predicted plumage visual contrast, especially for females. Our study therefore demonstrates the important role of non-sexually selected traits, viewing conditions and bird vision, in shaping avian plumage contrast.

The selective factors that shape phenotypic diversity in prey communities with aposematic animals are diverse and coincide with similar diversity in the strength of underlying secondary defences. However, quantitative assessments of colour pattern variation and the strength of chemical defences in assemblages of aposematic species are lacking. We quantified colour pattern diversity using quantitative colour pattern analysis (QCPA) in 13 dorid nudibranch species (Infraorder: Doridoidei) that varied in the strength of their chemical defences. We accounted for the physiological properties of a potential predator’s visual system (a triggerfish, Rhinecanthus aculeatus ) and modelled the appearance of nudibranchs from multiple viewing distances (2 and 10 cm). We identified distinct colour pattern properties associated with the presence and strength of chemical defences. Specifically, increases in chemical defences indicated increases in colour pattern boldness (i.e. visual contrast elicited via either or potentially coinciding chromatic, achromatic and/or spatial contrast). Colour patterns were also less variable among species with chemical defences when compared to undefended species. Our results indicate correlations between secondary defences and diverse, bold colouration while showing that chemical defences coincide with decreased colour pattern variability among species. Our study suggests that complex spatiochromatic properties of colour patterns perceived by potential predators can be used to make inferences on the presence and strength of chemical defences.

Sexual selection can result in extreme development of multimodal mate-attracting traits, including complex constructions. Male Great Bowerbirds build bowers for attracting females. Bowers contain a thatched twig tunnel (avenue) opening onto 2 courts covered with decorations. Males displaying on a court are seen by a female from within the avenue. She sees and hears displays through the avenue entrance but can only see the male's head and objects in his bill as it passes repeatedly across the entrance. Because the bower may affect the auditory as well as the visual parts of the multimodal male display we investigated bower acoustic properties by playing standard sounds from multiple court positions, recording the resulting sounds at the female's head position within the avenue. Bower geometry results in a limited zone at the avenue entrance where his vocalisations can be heard with maximum intensity; this corresponds to his typical display position. Experiments show that court decorations increase the intensity of some frequencies and reduce the intensity of others. Bower structure simultaneously affects both visual and auditory male display components and could be important in sexual selection. It is important to consider more than 1 sensory mode, especially in the context of built signaling structures. Sexual selection can result in complex traits. Bowerbirds construct and decorate bowers which are used to attract and hold a female's attention for mating while she watches from within the bower tunnel. Bowers allow a complex visual display but the structure also affects the sounds reaching the female. We showed that the structure limits the locations from which a male can transmit the loudest sounds to the female and also affects the pitch structure.

Survival games can be described as video games where the player searches for food and treasures, while avoiding obstacles and hostile attacks. Ms.Pac-Man and Minecraft are two well-known examples. Currently there are AI models that outperform human players at Ms.Pac-Man, while AI models playing Minecraft above the human level has been a long-standing challenge. This paper concerns what we call pure survival games, which take place in previously unseen worlds containing only food, water, and obstacles. The challenge of the player is to navigate and survive in those worlds by continuously finding resources and avoiding obstacles. Arguably, animals need to master physical analogues of pure survival games in order to survive and reproduce. Here we begin to explore human and machine performance on pure survival games. We define two games called the Grid game and the Terrain game and two corresponding AI agents based on deep reinforcement learning: the Grid agent and the Terrain agent. We explore to what extent these agents can match human performance and how their performance is affected by variations in their perception, memory, and reward models. We find that (1) the Terrain agent performs above human level, while the Grid agent performs below human level; (2) the smell, touch, and interoception models contribute significantly to the performance of the Grid agent; (3) the memory model contributes significantly to the performance of the Grid agent; and (4) the performance of the Grid agent is relatively stable under three quite different reward signals, including one that rewards survival and nothing else.

We present an AI-based ecosystem simulator that uses three-dimensional models of the terrain and animal models controlled by deep reinforcement learning. The simulations take place in a game engine environment, which enables continuous visual observation of the ecosystem model. The terrain models are generated from geographic data with altitudes and land cover type. The animal models combine three-dimensional conformation models with animation schemes and decision-making mechanisms trained with deep reinforcement learning in increasingly complex environments (curriculum learning). We show how AI tools of this kind can be used for modeling the development of specific ecosystems with and without different forms of economic activities. In particular, we show how they might be used for modeling local biodiversity effects of land cover change, exploitation of natural resources, pollution, invasive species, and climate change.

Pottery is a quintessential indicator of human cultural dynamics. Cultural alignment of behavioral repertoires and artifacts has been considered to rest upon two distinct dynamics: selective transmission of information and culture-specific biased transformation. In a cross-cultural field experiment, we tested whether community-specific morphological features of ceramic vessels would arise when the same unfamiliar shapes were reproduced by professional potters from three different communities who threw vessels using wheels. We analyzed the details of the underlying morphogenesis development of vessels in wheel throwing. When expert potters from three different communities of practice were instructed to faithfully reproduce common unfamiliar model shapes that were not parts of the daily repertoires, the morphometric variation in the final shape was not random; rather, different potters produced vessels with more morphometric variation among than within communities, indicating the presence of community-specific deviations of morphological features of vessels. Furthermore, this was found both in the final shape and in the underlying process of morphogenesis; there was more variation in the morphogenetic path among than within communities. These results suggest that the morphological features of ceramic vessels produced by potters reliably and nonrandomly diverge among different communities. The present study provides empirical evidence that collective alignment of morphological features of ceramic vessels can arise from the community-specific habits of fashioning clay.

The "escape and radiate" hypothesis predicts that once species have evolved aposematism, defended species can utilize more visually diverse visual backgrounds as they "escape" the need to be well camouflaged. This enables species to explore new ecological niches, resulting in increased diversification rates. To test this hypothesis "escape" component, we examined whether the background habitats of 12 nudibranch mollusk species differed among species depending on the presence and strength of chemical defenses. We obtained a rich array of color pattern statistics using quantitative color pattern analysis to analyze backgrounds viewed through the eyes of a potential predator (triggerfish, Rhinecanthus aculeatus). Color pattern analysis was done at viewing distances simulating an escalating predation sequence. We identified 4 latent factors comprising 17 noncorrelated color pattern parameters, which captured the among-species variability associated with differences in chemical defenses. We found that chemically defended species, indeed, were found on visually distinct backgrounds with increased color and luminance contrast, independent of viewing distance. However, we found no evidence for increased among-species background diversity coinciding with the presence and strength of chemical defenses. Our results agree with the "escape and radiate" hypothesis, suggesting that potent chemical defenses in Dorid nudibranchs coincide with spatiochromatic differences of visual background habitats perceived by potential predators.

Coloration facilitates evolutionary investigations in nature because the interaction between genotype, phenotype and environment is relatively accessible. In a landmark set of studies, Endler addressed this complexity by demonstrating that the evolution of male Trinidadian guppy coloration is shaped by the local balance between selection for mate attractiveness versus crypsis. This became a textbook paradigm for how antagonistic selective pressures may determine evolutionary trajectories in nature. However, recent studies have challenged the generality of this paradigm. Here, we respond to these challenges by reviewing five important yet underappreciated factors that contribute to colour pattern evolution: (i) among-population variation in female preference and correlated variation in male coloration, (ii) differences in how predators versus conspecifics view males, (iii) biased assessment of pigmentary versus structural coloration, (iv) the importance of accounting for multi-species predator communities, and (v) the importance of considering the multivariate genetic architecture and multivariate context of selection and how sexual selection encourages polymorphic divergence. We elaborate these issues using two challenging papers. Our purpose is not to criticize but to point out the potential pitfalls in colour research and to emphasize the depth of consideration necessary for testing evolutionary hypotheses using complex multi-trait phenotypes such as guppy colour patterns.

Diurnal biting flies are strongly attracted to blue objects. This behaviour is widely exploited for fly control, but its functional significance is debated. It is hypothesized that blue objects resemble animal hosts; blue surfaces resemble shaded resting places; and blue attraction is a by-product of attraction to polarized light. We computed the fly photoreceptor signals elicited by a large sample of leaf and animal integument reflectance spectra, viewed under open/cloudy illumination and under woodland shade. We then trained artificial neural networks (ANNs) to distinguish animals from leaf backgrounds, and shaded from unshaded surfaces, in order to find the optimal means of doing so based upon the sensory information available to a fly. After training, we challenged ANNs to classify blue objects used in fly control. Trained ANNs could make both discriminations with high accuracy. They discriminated animals from leaves based upon blue–green photoreceptor opponency and commonly misclassified blue objects as animals. Meanwhile, they discriminated shaded from unshaded stimuli using achromatic cues and never misclassified blue objects as shaded. We conclude that blue–green opponency is the most effective means of discriminating animals from leaf backgrounds using a fly's sensory information, and that blue objects resemble animal hosts through such mechanisms.

In a variety of aposematic species, the conspicuousness of an individual's warning signal and the quantity of its chemical defence are positively correlated. This apparent honest signalling is predicted by resource competition models which assume that the production and maintenance of aposematic defences compete for access to antioxidant molecules that have dual functions as pigments and in protecting against oxidative damage. To test for such trade-offs, we raised monarch butterflies ( Danaus plexippus ) on different species of their milkweed host plants (Apocynaceae) that vary in quantities of cardenolides to test whether (i) the sequestration of cardenolides as a secondary defence is associated with costs in the form of oxidative lipid damage and reduced antioxidant defences; and (ii) lower oxidative state is associated with a reduced capacity to produce aposematic displays. In male monarchs conspicuousness was explained by an interaction between oxidative damage and sequestration: males with high levels of oxidative damage became less conspicuous with increased sequestration of cardenolides, whereas those with low oxidative damage became more conspicuous with increased levels of cardenolides. There was no significant effect of oxidative damage or concentration of sequestered cardenolides on female conspicuousness. Our results demonstrate a physiological linkage between the production of coloration and oxidative state, and differential costs of sequestration and signalling in monarch butterflies.

Aposematic signals visually advertise underlying anti-predatory defences in many species. They should be detectable (e.g. contrasting against the background) and bold (e.g. using internal pattern contrast) to enhance predator recognition, learning and memorization. However, the signalling function of aposematic colour patterns may be distance-dependent: signals may be undetectable from a distance to reduce increased attacks from naïve predators but bold when viewed up close. Using quantitative colour pattern analysis, we quantified the chromatic and achromatic detectability and boldness of colour patterns in 13 nudibranch species with variable strength of chemical defences in terms of unpalatability and toxicity, approximating the visual perception of a triggerfish (Rhinecanthus aculeatus) across a predation sequence (detection to subjugation). When viewed from an ecologically relevant distance of 30 cm, there were no differences in detectability and boldness between well-defended and undefended species. However, when viewed at closer distances (less than 30 cm), well-defended species were more detectable and bolder than undefended species. As distance increased, detectability decreased more significantly than boldness for defended species. For undefended species, boldness and detectability remained comparatively consistent, regardless of viewing distance. We provide evidence for distance-dependent signalling in aposematic nudibranchs and highlight the importance of distinguishing signal detectability from boldness in studies of aposematism.

The rarity of tool manufacture in wild parrots is surprising because they share key life-history traits with advanced tool-using species, including large brains, complex sociality and prolonged parental care. When it does occur, tool manufacture in parrots tends to be innovative, spontaneous and individually variable, but most cases have been in captivity. In the wild, only palm cockatoos (Probosciger aterrimus) have been observed using tools regularly. However, they are unusual because they use tools to enhance their displays rather than for foraging or self-maintenance. Males in northern Australia make two types of tool from sticks and seed pods, which they tap rhythmically against a tree during display. We analysed 256 sound tools retrieved from 70 display trees. Drumsticks (89% of tools) were used more often than seed pod tools; most males manufactured only drumsticks, but some made both types. Individual males differed significantly in the design of their drumsticks including the length, width and mass but we found no evidence that neighbours copied each other. We discuss the highly individualized preferences for sound tool design in context of the behavioural predispositions behind the rarity of tool manufacture in wild parrots.

The string-pulling paradigm is an approach commonly used in the study of animal cognition to investigate problem-solving abilities. This test involves an out-of-reach reward that can only be acquired through pulling a string. Australian magpies ( Gymnorhina tibicen tyrannica ) can solve cognitive tasks requiring associative and reversal learning, spatial memory, and inhibitory control. Nonetheless, whether magpies can pass a test of means-end understanding—the string-pull test—is unclear. We tested wild magpies on a string-pulling task with five configurations, including a long loose string that required several pulls to obtain food, a short string that only required a single pull to complete, and two short strings—one intact and connected to the food reward; the other broken and unable to yield food when pulled. Of the 11 magpies tested, none solved the long string task at first presentation. Two naïve birds and three, subsequently trained, birds succeeded at the short string. Once proficient at the short string, four of these five birds then solved the long string task; in addition, three learned to choose a functional, intact string over a non-functional, broken string. Overall, these observations provide evidence that Australian magpies have the ability to solve string-pulling tasks but it remains unclear whether they do so by trial-and-error or if they possess means-end understanding. Significance statement We found, for the first time, that wild, free-ranging Australian magpies can learn to solve different versions of a string-pulling task; a version with a short string, a long string, and two strings next to each other where one of them was broken and non-functional (broken string task). Some magpies spontaneously solved the task with the short string, where other magpies required training, and magpies only managed to solve the task with the long string after they had succeeded on the short string task. Furthermore, some of the magpies solved the broken string task but required a high number of trials to do so. Overall, our findings show that Australian magpies have the ability to solve string-pulling tasks but it remains unclear whether they do so by trial-and-error or if they understand causal relations between objects.

. We present an ecosystem simulator with three-dimensional models of topography and terrain. The simulator includes animal models that are individually controlled by deep reinforcement learning. The simulations take place in a game engine environment, which enables continuous observation of the ecosystem from any camera angle. The topographic models are generated from geographic data with altitudes and land cover type. The animal models combine three-dimensional confor-mation models and animation schemes with decision-making mechanisms trained with deep reinforcement learning in increasingly complex environments (curriculum learning). We show how this framework can be used for simulating the development of specific ecosystems with or without different forms of human intervention. In particular, we show how it can be used for studying local biodiversity effects of land use change, exploitation of natural resources, pollution, invasive species, and climate change.

ABSTRACT Animals use colour vision in a range of behaviours. Visual performance is limited by thresholds, which are set by noise in photoreceptors and subsequent neural processing. The receptor noise limited (RNL) model of colour discrimination is widely used for modelling colour vision and accounts well for experimental data from many species. In one of the most comprehensive tests yet of colour discrimination in a non-human species, we used Ishihara-style stimulus patterns to examine thresholds for 21 directions at five locations in colour space for the fish Rhinecanthus aculeatus. Thresholds matched RNL model predictions most closely for stimuli near the achromatic point, but exceeded predictions (indicating a decline in sensitivity) with distance from this point. Thresholds were also usually higher for saturation than for hue differences. These changes in colour threshold with colour space location and direction may give insight into photoreceptor non-linearities and post-receptoral mechanisms of colour vision in fish. Our results highlight the need for a cautious interpretation of the RNL model – especially for modelling colours that differ from one another in saturation (rather than hue), and for highly saturated colours distant from the achromatic point in colour space.

Maintenance of variation in aposematic traits within and among populations is paradoxical because aposematic species are normally under positive frequency-dependent predation (PFD), which is expected to erode variation. Aposematic traits can evolve in an ecological context where aposematic traits are simultaneously under mate choice. Here, we examine how the mate preference intensity affects the permissiveness of polymorphism in sexually selected aposematic traits under different PFD regimes. We use the haploid version of the classical sexual selection model and show that strong mate preferences can substantially increase the permissiveness of polymorphism in aposematic traits under different PFD regimes. The Fisher process can interact with PFD, and their interaction can promote the maintenance of polymorphism within populations when mate preferences are strong. We show that the same selective conditions that promote the maintenance of polymorphism within populations reduce the likelihood of divergence in aposematic traits among populations.

Deimatic behaviours, also referred to as startle behaviours, are used against predators and rivals. Although many are spectacular, their proximate and ultimate causes remain unclear. In this review we aim to synthesise what is known about deimatic behaviour and identify knowledge gaps. We propose a working hypothesis for deimatic behaviour, and discuss the available evidence for the evolution, ontogeny, causation, and survival value of deimatic behaviour using Tinbergen's Four Questions as a framework. Our overarching aim is to direct future research by suggesting ways to address the most pressing questions in this field.

Male guppies (Poecilia reticulata) have multiple colored spots and perform courtship displays near the edges of streams in Trinidad in shallow water flowing through rainforest. Depending upon the orientation of the pair, the female sees the male displays against gravel or other stream bed substrates or against the spacelight-the roughly uniform light coming from the water column away from the bank. We observed courting pairs in two adjacent natural streams and noted the directions of each male display. We found that the female sees the male more often against spacelight than against gravel when females either faced the spacelight from the opposite bank or from downstream, or both. Visual modelling using natural substrate reflectances and field light measurements showed higher chromatic contrast of males against spacelight than against substrates independent of the two ambient light environments used during displays, but achromatic contrast depended upon the ambient light habitat. This suggests that courtship involves both chromatic and achromatic contrast. We conclude that the orientation of courting pairs and the ambient light spectrum should be accounted for in studies of mate choice, because the visual background and light affect visibility, and these differ with orientation.

Great bowerbird males build bowers for attracting females for mating. Bowers consist of a thatched twig tunnel (avenue) which opens onto two flat courts covered with objects. Male displays on a court are seen by a female from within the avenue. She sees and hears displays through the avenue entrance but can only see the male’s head and objects in his bill as it passes repeatedly across the entrance. We investigated bower acoustic properties by playing standard sounds from multiple court positions and recorded the resulting sounds at the female’s typical avenue head position within the avenue. Bower geometry significantly affects both his acoustic and visual display components and physically synchronizes them as he repeatedly moves in and out of the female’s view. Consequently, complex neural circuitry is unnecessary for linking sound to vision. Experimentally removing bower objects shows that objects significantly increase higher frequencies, hence bandwidth and loudness received inside the avenue. Great Bowerbird bowers produce a synchronized multimodal signal to females which may increase male attractiveness more than if a single sensory mode were used. This multimodal signal is unusual in that it is constructed rather than being part of the body and synchronized as a result of physics rather than neurons. Summary Statement Bowerbird bower geometry jointly effects both visual and auditory signal components and synchronizes them without the need for additional neural circuitry.

Ecosystem models can be used for understanding general phenomena of evolution, ecology, and ethology. They can also be used for analyzing and predicting the ecological consequences of human activities on specific ecosystems, e.g., the effects of agriculture, forestry, construction, hunting, and fishing. We argue that powerful ecosystem models need to include reasonable models of the physical environment and of animal behavior. We also argue that several well-known ecosystem models are unsatisfactory in this regard. Then we present the open-source ecosystem simulator Ecotwin, which is built on top of the game engine Unity. To model a specific ecosystem in Ecotwin, we first generate a 3D Unity model of the physical environment, based on topographic or bathymetric data. Then we insert digital 3D models of the organisms of interest into the environment model. Each organism is equipped with a genome and capable of sexual or asexual reproduction. An organism dies if it runs out of some vital resource or reaches its maximum age. The animal models are equipped with behavioral models that include sensors, actions, reward signals, and mechanisms of learning and decision-making. Finally, we illustrate how Ecotwin works by building and running one terrestrial and one marine ecosystem model.