Optimizing colour for camouflage and visibility using deep learning: the effects of the environment and the observer's visual system

Through an animal's eye: the implications of diverse sensory systems in scientific experimentation

'Accounting for the sensory abilities of animals is critical in experimental design.' No researcher would disagree with this statement, yet it is often the case that we inadvertently fall for anthropocentric biases and use ourselves as the reference point. This paper discusses the risks of adopting an anthropocentric view when working with non-human animals, and the unintended consequences this has on our experimental designs and results. To this aim, we provide general examples of anthropocentric bias from different fields of animal research, with a particular focus on animal cognition and behaviour, and lay out the potential consequences of adopting a human-based perspective. Knowledge of the sensory abilities, both in terms of similarities to humans and peculiarities of the investigated species, is crucial to ensure solid conclusions. A more careful consideration of the diverse sensory systems of animals would improve many scientific fields and enhance animal welfare in the laboratory.

Adapting genetic algorithms for artificial evolution of visual patterns under selection from wild predators

Camouflage is a widespread and well-studied anti-predator strategy, yet identifying which patterns provide optimal protection in any given scenario remains challenging. Besides the virtually limitless combinations of colours and patterns available to prey, selection for camouflage strategies will depend on complex interactions between prey appearance, background properties and predator traits, across repeated encounters between co-evolving predators and prey. Experiments in artificial evolution, pairing psychophysics detection tasks with genetic algorithms, offer a promising way to tackle this complexity, but sophisticated genetic algorithms have so far been restricted to screen-based experiments. Here, we present methods to test the evolution of colour patterns on physical prey items, under selection from wild predators in the field. Our techniques expand on a recently-developed open-access pattern generation and genetic algorithm framework, modified to operate alongside artificial predation experiments. In this system, predators freely interact with prey, and the order of attack determines the survival and reproduction of prey patterns into future generations. We demonstrate the feasibility of these methods with a case study, in which free-flying birds feed on artificial prey deployed in semi-natural conditions, against backgrounds differing in three-dimensional complexity. Wild predators reliably participated in this experiment, foraging for 11 to 16 generations of artificial prey and encountering a total of 1,296 evolved prey items. Changes in prey pattern across generations indicated improvements in several metrics of similarity to the background, and greater edge disruption, although effect sizes were relatively small. Computer-based replicates of these trials, with human volunteers, highlighted the importance of starting population parameters for subsequent evolution, a key consideration when applying these methods. Ultimately, these methods provide pathways for integrating complex genetic algorithms into more naturalistic predation trials. Customisable open-access tools should facilitate application of these tools to investigate a wide range of visual pattern types in more ecologically-relevant contexts.

A Review of Cervidae Visual Ecology

This review examines the visual systems of cervids in relation to their ability to meet their ecological needs and how their visual systems are specialized for particular tasks. Cervidae encompasses a diverse group of mammals that serve as important ecological drivers within their ecosystems. Despite evidence of highly specialized visual systems, a large portion of cervid research ignores or fails to consider the realities of cervid vision as it relates to their ecology. Failure to account for an animal's visual ecology during research can lead to unintentional biases and uninformed conclusions regarding the decision making and behaviors for a species or population. Our review addresses core behaviors and their interrelationship with cervid visual characteristics. Historically, the study of cervid visual characteristics has been restricted to specific areas of inquiry such as color vision and contains limited integration into broader ecological and behavioral research. The purpose of our review is to bridge these gaps by offering a comprehensive review of cervid visual ecology that emphasizes the interplay between the visual adaptations of cervids and their interactions with habitats and other species. Ultimately, a better understanding of cervid visual ecology allows researchers to gain deeper insights into their behavior and ecology, providing critical information for conservation and management efforts.

Little information loss with red-green color deficient vision in natural environments

Inherited color vision deficiency affects red-green discrimination in about one in twelve men from European populations. Its effects have been studied mainly in primitive foraging but also in detecting blushing and breaking camouflage. Yet there is no obvious relationship between these specific tasks and vision in the real world. The aim here was to quantify the impact of color vision deficiency by estimating computationally the information available to observers about colored surfaces in natural scenes. With representative independent sets of 50 and 100 hyperspectral images, estimated information was found to be only a little less in red-green color vision deficiency than in normal trichromacy. Colorimetric analyses revealed the importance of large lightness variations within scenes, small redness-greenness variations, and uneven frequencies of different colored surfaces. While red-green color vision deficiency poses challenges in some tasks, it has much less effect on gaining information from natural environments.

Varying benefits of generalist and specialist camouflage in two versus four background environments

Background-matching camouflage is a well-established strategy to reduce detection, but implementing this on heterogeneous backgrounds is challenging. For prey with fixed color patterns, solutions include specializing on a particular visual microhabitat, or adopting a compromise or generalist appearance, matching multiple backgrounds less well. Existing studies suggest both approaches can succeed, but most consider relatively simple scenarios, where artificial prey appear against two backgrounds differing in a single visual characteristic. Here, we used computer-based search tasks with human participants to test the relative benefits of specializing and generalizing for complex targets, displayed on either two or four types of naturalistic backgrounds. Across two background types, specialization was beneficial on average. However, the success of this strategy varied with search duration, such that generalist targets could outperform specialists over short search durations due to the presence of poorly matched specialists. Over longer searches, the remaining well-matched specialists had greater success than generalists, leading to an overall benefit of specialization at longer search durations. Against four different backgrounds, the initial cost to specialization was greater, so specialists and generalists ultimately experienced similar survival. Generalists performed better when their patterning was a compromise between backgrounds that were more similar to each other than when backgrounds were more different, with similarity in luminance more relevant than pattern differences. Time dependence in the relative success of these strategies suggests that predator search behavior may affect optimal camouflage in real-world situations.

The Colours of Octopus: Using Spectral Data to Measure Octopus Camouflage

No animal can so effectively camouflage in such a wide range of environments as the octopus. Thanks to their highly malleable skin, they are capable of adapting their body patterns to the brightness and texture of their immediate environment, and they often seemingly match the colour of background objects. However, octopuses are colour-blind as their eyes have only one type of visual pigment. Therefore, chromatophores in their skin are likely to respond to changes in brightness, not chromaticity. To determine whether octopuses actually match background colours, we used a SpectraScan® PR-655 spectroradiometer to measure the reflectance spectra of Octopus tetricus skin in captivity. The spectra were compared with those of green algae, brown algae, and sponges—all of these being colourful objects commonly found in the octopus’s natural environment. Even though we show that octopuses change both lightness and chromaticity, allowing them to potentially camouflage in a wide range of backgrounds in an effective manner, the overall octopus colours did not reach the same level of saturation compared to some background objects. Spectra were then modelled under the visual systems of four potential octopus predators: one dichromatic fish (Heller’s barracuda), two trichromatic fish (blue-spotted stingray and two-spotted red snapper), and one tetrachromatic bird (wedge-tailed shearwater). We show that octopuses are able to match certain background colours for some visual systems. How a colour-blind animal is capable of colour-matching is still unknown.

Color vision deficiencies and camouflage: a comparative study between normal and CVD observers

There is a belief that observers with color vision deficiencies (CVD) perform better in detecting camouflaged objects than normal observers. Some studies have concluded contradictory findings when studying the performance of normal and CVD observers in the camouflage detection tasks in different conditions. This work presents a literature review on this topic, dividing it into three different and contradictory types of results: better performance for CVD, for normal observers, or same performance. Besides, two psychophysical experiments have been designed and carried out in a calibrated computer monitor on both normal and CVD human observers to measure the searching times of the different types of observers needed to find camouflaged stimuli in two different types of stimuli. Results show the trend that, in our experimental conditions, normal observers need shorter searching times than CVD observers in finding camouflaged stimuli both in images of natural scenes and in images with synthetic stimuli.

Generalist camouflage can be more successful than microhabitat specialisation in natural environments

BACKGROUND

Crypsis by background-matching is a critical form of anti-predator defence for animals exposed to visual predators, but achieving effective camouflage in patchy and variable natural environments is not straightforward. To cope with heterogeneous backgrounds, animals could either specialise on particular microhabitat patches, appearing cryptic in some areas but mismatching others, or adopt a compromise strategy, providing partial matching across different patch types. Existing studies have tested the effectiveness of compromise strategies in only a limited set of circumstances, primarily with small targets varying in pattern, and usually in screen-based tasks. Here, we measured the detection risk associated with different background-matching strategies for relatively large targets, with human observers searching for them in natural scenes, and focusing on colour. Model prey were designed to either 'specialise' on the colour of common microhabitat patches, or 'generalise' by matching the average colour of the whole visual scenes.

RESULTS

In both the field and an equivalent online computer-based search task, targets adopting the generalist strategy were more successful in evading detection than those matching microhabitat patches. This advantage occurred because, across all possible locations in these experiments, targets were typically viewed against a patchwork of different microhabitat areas; the putatively generalist targets were thus more similar on average to their various immediate surroundings than were the specialists.

CONCLUSIONS

Demonstrating close agreement between the results of field and online search experiments provides useful validation of online citizen science methods commonly used to test principles of camouflage, at least for human observers. In finding a survival benefit to matching the average colour of the visual scenes in our chosen environment, our results highlight the importance of relative scales in determining optimal camouflage strategies, and suggest how compromise coloration can succeed in nature.

Conspicuousness, phylogenetic structure, and origins of Müllerian mimicry in 4000 lycid beetles from all zoogeographic regions

Biologists have reported on the chemical defences and the phenetic similarity of net-winged beetles (Coleoptera: Lycidae) and their co-mimics. Nevertheless, our knowledge has remained fragmental, and the evolution of mimetic patterns has not been studied in the phylogenetic context. We illustrate the general appearance of ~ 600 lycid species and ~ 200 co-mimics and their distribution. Further, we assemble the phylogeny using the transcriptomic backbone and ~ 570 species. Using phylogenetic information, we closely scrutinise the relationships among aposematically coloured species, the worldwide diversity, and the distribution of aposematic patterns. The emitted visual signals differ in conspicuousness. The uniform coloured dorsum is ancestral and was followed by the evolution of bicoloured forms. The mottled patterns, i.e. fasciate, striate, punctate, and reticulate, originated later in the course of evolution. The highest number of sympatrically occurring patterns was recovered in New Guinea and the Andean mountain ecosystems (the areas of the highest abundance), and in continental South East Asia (an area of moderate abundance but high in phylogenetic diversity). Consequently, a large number of co-existing aposematic patterns in a single region and/or locality is the rule, in contrast with the theoretical prediction, and predators do not face a simple model-like choice but cope with complex mimetic communities. Lycids display an ancestral aposematic signal even though they sympatrically occur with differently coloured unprofitable relatives. We show that the highly conspicuous patterns evolve within communities predominantly formed by less conspicuous Müllerian mimics and, and often only a single species displays a novel pattern. Our work is a forerunner to the detailed research into the aposematic signalling of net-winged beetles.

The effect of pelage, background, and distance on predator detection and the evolution of primate color vision

Primates' predators, such as carnivore mammals, usually rely on camouflage to increase proximity to prey and start a predatory attempt. Camouflage depends not only on the color pattern presented by a predator's pelage but also on the background scene in which the hunting takes place. Another factor that influences camouflage effectiveness is prey's color vision since a given camouflage strategy might not work for all visual phenotypes. Still, little research has been made on the effect of primate visual phenotype on predator detection. Here, we investigate the effects of natural pelages, background scenarios, visual phenotypes, and detection distances on predator detection. We used photographs of taxidermized carnivores (ocelots, cougars, and lesser grisons) as detection stimuli, taken in three different natural scenarios (forest, savanna, and grassland), and at two viewing distances (near and far). On a touchscreen monitor, sets of four images (only one containing a hidden animal) were randomly presented to 39 human males (19 dichromats and 20 trichromats). We found that trichromats, when compared to dichromats, present a lower latency and a higher accuracy of carnivore detection for some conditions tested. We also found that pelage color, background scenario, and detection distance interact to influence the effectiveness of camouflage. Our results suggest that trichromacy might be even more advantageous for carnivore detection than thought before, since it facilitates detection of mammals with diverse pelage colorations, in environments with different phytophysiognomies, and at longer distances. We also propose that the higher rates of dichromacy found in modern human societies could have resulted from a relaxation in predation.

The Camouflage Machine: Optimizing protective coloration using deep learning with genetic algorithms

Evolutionary biologists frequently wish to measure the fitness of alternative phenotypes using behavioral experiments. However, many phenotypes are complex. One example is coloration: camouflage aims to make detection harder, while conspicuous signals (e.g., for warning or mate attraction) require the opposite. Identifying the hardest and easiest to find patterns is essential for understanding the evolutionary forces that shape protective coloration, but the parameter space of potential patterns (colored visual textures) is vast, limiting previous empirical studies to a narrow range of phenotypes. Here, we demonstrate how deep learning combined with genetic algorithms can be used to augment behavioral experiments, identifying both the best camouflage and the most conspicuous signal(s) from an arbitrarily vast array of patterns. To show the generality of our approach, we do so for both trichromatic (e.g., human) and dichromatic (e.g., typical mammalian) visual systems, in two different habitats. The patterns identified were validated using human participants; those identified as the best for camouflage were significantly harder to find than a tried-and-tested military design, while those identified as most conspicuous were significantly easier to find than other patterns. More generally, our method, dubbed the "Camouflage Machine," will be a useful tool for identifying the optimal phenotype in high dimensional state spaces.

Ecological, behavioral, and phylogenetic influences on the evolution of dorsal color pattern in geckos

The dorsal surfaces of many taxonomic groups often feature repetitive pattern elements consisting of stripes, spots, or bands. Here, we investigate how distinct categories of camouflage pattern work by relating them to ecological and behavioral traits in 439 species of gecko. We use phylogenetic comparative methods to test outstanding hypotheses based on camouflage theory and research in other taxa. We found that bands are associated with nocturnal activity, suggesting bands provide effective camouflage for motionless geckos resting in refugia during the day. A predicted association between stripes and diurnal activity was not supported, suggesting that stripes do not work via dazzle camouflage mechanisms in geckos. This, along with a lack of support for our prediction that plain patterning should be associated with open habitats, suggests that similar camouflage patterns do not work in consistent ways across taxa. We also found that plain and striped lineages frequently switched between using open or closed habitats, whereas spotted lineages rarely transitioned. This suggests that pattern categories differ in how specialized or generalized their camouflage is. This result has ramifications for theory on how camouflage compromises to background heterogeneity and how camouflage pattern might influence evolutionary trajectories.