Color vision models: Some simulations, a general n-dimensional model, and the colourvision R package

Automated workflows using Quantitative Colour Pattern Analysis (QCPA): a guide to batch processing and downstream data analysis

Animal and plant colouration presents a striking dimension of phenotypic variation, the study of which has driven general advances in ecology, evolution, and animal behaviour. Quantitative Colour Pattern Analysis (QCPA) is a dynamic framework for analysing colour patterns through the eyes of non-human observers. However, its extensive array of user-defined image processing and analysis tools means image analysis is often time-consuming. This hinders the full use of analytical power provided by QCPA and its application to large datasets. Here, we offer a robust and comprehensive batch script, allowing users to automate many QCPA workflows. We also provide a complimentary set of useful R scripts for downstream data extraction and analysis. The presented batch processing extension will empower users to further utilise the analytical power of QCPA and facilitate the development of customised semi-automated workflows. Such quantitatively scaled workflows are crucial for exploring colour pattern spaces and developing ever-richer frameworks for analysing organismal colouration accounting for visual perception in animals other than humans. These advances will, in turn, facilitate testing hypotheses on the function and evolution of vision and signals at quantitative and qualitative scales, which are otherwise computationally unfeasible.

Supplementary Information

The online version contains supplementary material available at 10.1007/s10682-024-10291-7.

Analysing biological colour patterns from digital images: An introduction to the current toolbox

Understanding the numerous roles that colouration serves in the natural world has remained a central focus in many evolutionary and ecological studies. However, to accurately characterise and then compare colours or patterns among individuals or species has been historically challenging. In recent years, there have been a myriad of new resources developed that allow researchers to characterise biological colours and patterns, specifically from digital imagery. However, each resource has its own strengths and weaknesses, answers a specific question and requires a detailed understanding of how it functions to be used properly. These nuances can make navigating this emerging field rather difficult. Herein, we evaluate several new techniques for analysing biological colouration, with a specific focus on digital images. First, we introduce fundamental background knowledge about light and perception to be considered when designing and implementing a study of colouration. We then show how numerous modifications can be made to images to ensure consistent formatting prior to analysis. After, we describe many of the new image analysis approaches and their respective functions, highlighting the type of research questions that they can address. We demonstrate how these various techniques can be brought together to examine novel research questions and test specific hypotheses. Finally, we outline potential future directions in colour pattern studies. Our goal is to provide a starting point and pathway for researchers wanting to study biological colour patterns from digital imagery.

Dim-light colour vision in the facultatively nocturnal Asian giant honeybee, Apis dorsata

We discovered nocturnal colour vision in the Asian giant honeybee Apis dorsata-a facultatively nocturnal species-at mesopic light intensities, down to half-moon light levels (approx. 10-2 cd m-2). The visual threshold of nocturnality aligns with their reported nocturnal activity down to the same light levels. Nocturnal colour vision in A. dorsata is interesting because, despite being primarily diurnal, its colour vision capabilities extend into dim light, while the 'model' European honeybee Apis mellifera is reported to be colour-blind at twilight. By employing behavioural experiments with naturally nesting A. dorsata colonies, we show discrimination of the trained colour from other stimuli during the day, and significantly, even at night. Nocturnal colour vision in bees has so far only been reported in the obligately nocturnal carpenter bee Xylocopa tranquebarica. The discovery of colour vision in these two bee species, despite differences in the extent of their nocturnality and the limitations of their apposition compound eye optics, opens avenues for future studies on visual adaptations for dim-light colour vision, their role in pollination of flowers at night, and the effect of light pollution on nocturnal activity in A. dorsata, a ubiquitous pollinator in natural, agricultural and urban habitats in the Asian tropics and sub-tropics.

Caution with colour calculations: spectral purity is a poor descriptor of flower colour visibility

BACKGROUND

The colours of flowers are of key interest to plant and pollination biologists. An increasing number of studies have investigated the importance of saturation of flower colours (often called 'spectral purity' or 'chroma') for visibility to pollinators, but the conceptual, physiological and behavioural foundations for these metrics as well as the calculations used rest on slender foundations.

METHODS

We discuss the caveats of colour attributes that are derived from human perception, and in particular spectral purity and chroma, as variables in flower colour analysis. We re-analysed seven published datasets encompassing 774 measured reflectance spectra to test for correlations between colour contrast, spectral purity and chroma.

MAIN FINDINGS AND CONCLUSIONS

We identify several concerns with common calculation procedures in animal colour spaces. Studies on animal colour vision provide no ground to assume that any pollinator perceives (or responds to) saturation, chroma or spectral purity in the way humans do. A re-analysis of published datasets revealed that values for colour contrast between flowers and their background are highly correlated with measures for spectral purity and chroma, which invalidates treating these factors as independent variables as is currently commonplace. Strikingly, spectral purity and chroma - both of which are metrics for saturation and are often used synonymously - are not correlated at all. We conclude that alternative, behaviourally validated metrics for the visibility of flowers to pollinators, such as colour contrast and achromatic contrast, are better in understanding the role of flower colour in plant-pollinator signalling.

Color discrimination thresholds vary throughout color space in a reef fish (Rhinecanthus aculeatus)

Animal use color 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 color discrimination is widely used for modelling color vision and accounts well for experimental data from many species. In one of the most comprehensive tests yet of color discrimination in a non-human species, we using Ishihara-style stimulus patterns to examine thresholds for 21 directions at five locations in color space for the fish Rhineacanthus aculeatus. Thresholds matched RNL model predictions most closely for stimuli near to the 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 color threshold with color space location and direction may give insight into photoreceptor non-linearities and post-receptoral mechanisms of color 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 especially for highly saturated colours distant from the achromatic point in colour space.

Flower Color Evolution and the Evidence of Pollinator-Mediated Selection

The evolution of floral traits in animal-pollinated plants involves the interaction between flowers as signal senders and pollinators as signal receivers. Flower colors are very diverse, effect pollinator attraction and flower foraging behavior, and are hypothesized to be shaped through pollinator-mediated selection. However, most of our current understanding of flower color evolution arises from variation between discrete color morphs and completed color shifts accompanying pollinator shifts, while evidence for pollinator-mediated selection on continuous variation in flower colors within populations is still scarce. In this review, we summarize experiments quantifying selection on continuous flower color variation in natural plant populations in the context of pollinator interactions. We found that evidence for significant pollinator-mediated selection is surprisingly limited among existing studies. We propose several possible explanations related to the complexity in the interaction between the colors of flowers and the sensory and cognitive abilities of pollinators as well as pollinator behavioral responses, on the one hand, and the distribution of variation in color phenotypes and fitness, on the other hand. We emphasize currently persisting weaknesses in experimental procedures, and provide some suggestions for how to improve methodology. In conclusion, we encourage future research to bring together plant and animal scientists to jointly forward our understanding of the mechanisms and circumstances of pollinator-mediated selection on flower color.

Body coloration as a dynamic signal during intrasexual communication in a cichlid fish

Background

Intrasexual competition over access to resources can lead to aggression between individuals. Because overt aggression, i.e. fights, can be costly for contestants, the communication of aggressive motivation prior to engagement in a physical fight is often mediated by conventional signals. Animals of various taxa, including fishes, display visual signals such as body coloration that can dynamically be adjusted depending on the individual’s motivation. Male individuals of the West African cichlid Pelvicachromis taeniatus express a yellow body coloration displayed during courtship but also in an intrasexual competition context.

Results

Within-individual variation in male yellow body coloration, as quantified with standardized digital photography and representation in a CIELab color space, was examined in a mating context by exposing males to a female and in a competitive intrasexual context, i.e. in a dyadic contest. Additionally, spectrometric reflectance measurements were taken to obtain color representations in a physiological color space based on spectral sensitivities of our model species. Exposure to females did not significantly affect male color expression. However, analysis of body coloration revealed a change in within-individual color intensity and colored area after interaction with a male competitor. In dominant males, extension of coloration was positively correlated with restrained aggression, i.e. displays, which in turn explained dominance established between the two contestants.

Conclusion

Body coloration in male P. taeniatus is a dynamic signal that is used in concert with display behavior in communication during intrasexual competition.

Ultraviolet vision aids the detection of nutrient-dense non-signaling plant foods

To expand our understanding of what tasks are particularly helped by UV vision and may justify the costs of focusing high-energy light onto the retina, we used an avian-vision multispectral camera to image diverse vegetated habitats in search of UV contrasts that differ markedly from visible-light contrasts. One UV contrast that stood out as very different from visible-light contrasts was that of nutrient-dense non-signaling plant foods (such as young leaves and immature fruits) against their natural backgrounds. From our images, we calculated color contrasts between 62+ species of such foods and mature foliage for the two predominant color vision systems of birds, UVS and VS. We also computationally generated images of what a generalized tetrachromat, unfiltered by oil droplets, would see, by developing a new methodology that uses constrained linear least squares to solve for optimal weighted combinations of avian camera filters to mimic new spectral sensitivities. In all visual systems, we found that nutrient-dense non-signaling plant foods presented a lower, often negative figure-ground contrast in the UV channels, and a higher, often positive figure-ground contrast in the visible channels. Although a zero contrast may sound unhelpful, it can actually enhance color contrast when compared in a color opponent system to other channels with nonzero contrasts. Here, low or negative UV contrasts markedly enhanced color contrasts. We propose that plants may struggle to evolve better UV crypsis since UV reflectance from vegetation is largely specular and thus highly dependent on object orientation, shape, and texture.

Behavioural thresholds of blue tit colour vision and the effect of background chromatic complexity

Vision is a vital attribute to foraging, navigation, mate selection and social signalling in animals, which often have a very different colour perception in comparison to humans. For understanding how animal colour perception works, vision models provide the smallest colour difference that animals of a given species are assumed to detect. To determine the just-noticeable-difference, or JND, vision models use Weber fractions that set discrimination thresholds of a stimulus compared to its background. However, although vision models are widely used, they rely on assumptions of Weber fractions since the exact fractions are unknown for most species. Here, we test; i) which Weber fractions in long-, middle- and shortwave (i.e. L, M, S) colour channels best describe the blue tit (Cyanistes caeruleus) colour discrimination, ii) how changes in hue of saturated colours and iii) chromatic background noise impair search behaviour in blue tits. We show that the behaviourally verified Weber fractions on achromatic backgrounds were L: 0.05, M: 0.03 and S: 0.03, indicating a high colour sensitivity. In contrast, on saturated chromatic backgrounds, the correct Weber fractions were considerably higher for L: 0.20, M: 0.17 and S: 0.15, indicating a less detailed colour perception. Chromatic complexity of backgrounds affected the longwave channel, while middle- and shortwave channels were mostly unaffected. We caution that using a vision model whereby colour discrimination is determined in achromatic viewing conditions, as they often are, can lead to misleading interpretations of biological interactions in natural - colourful - environments.

Decontextualized learning for interpretable hierarchical representations of visual patterns

Apart from discriminative modeling, the application of deep convolutional neural networks to basic research utilizing natural imaging data faces unique hurdles. Here, we present decontextualized hierarchical representation learning (DHRL), designed specifically to overcome these limitations. DHRL enables the broader use of small datasets, which are typical in most studies. It also captures spatial relationships between features, provides novel tools for investigating latent variables, and achieves state-of-the-art disentanglement scores on small datasets. DHRL is enabled by a novel preprocessing technique inspired by generative model chaining and an improved ladder network architecture and regularization scheme. More than an analytical tool, DHRL enables novel capabilities for virtual experiments performed directly on a latent representation, which may transform the way we perform investigations of natural image features, directly integrating analytical, empirical, and theoretical approaches.

Horse vision and obstacle visibility in horseracing

•

Using white, fluorescent yellow, or bright blue for fence markers would provide higher contrast for horses than the current orange markers.

•

Contrast is maximised across light and weather conditions by highly luminant whites or blues at the base of the fence (take-off board).

•

Fluorescent yellow has the greatest contrast against the main fence body (i.e. when used for midrail colour) across different light and weather conditions.

•

Horses jumped differently over white, fluorescent yellow, and bright blue marked fences compared to orange marked fences.

Visual information is key to how many animals interact with their environment, and much research has investigated how animals respond to colour and brightness information in the natural world. Understanding the visibility of features in anthropogenic environments, and how animals respond to these, is also important, not least for the welfare and safety of animals and the humans they co-exist with, but has received comparatively less attention. One area where this is particularly pertinent is animal sports such as horseracing. Here there is a need to understand how horses see and respond to obstacles, predominantly fences and hurdles, as this has implications for horse and rider safety, however obstacle appearance is currently designed to human perception. Using models of horse colour and luminance (perceived lightness) vision, we analysed the contrast of traditional orange markers currently used on fences from 11 UK racecourses, and compared this to potential alternative colours, while also investigating the effect of light and weather conditions on contrast. We found that for horses, orange has poor visibility and contrast against most surroundings. In comparison, yellow, blue, and white are more conspicuous, with the degree of relative contrast varying with vegetation or background type. Results were mostly consistent under different weather conditions and time of day, except for comparisons with the foreground turf in shade. We then tested the jump responses of racehorses to fences with orange, fluorescent yellow, bright blue, or white takeoff boards and midrails. Fence colour influenced both the angle of the jump and the distances jumped. Bright blue produced a larger angle of takeoff, and jumps over fluorescent yellow fences had shorter landing distances compared to orange, with bright blue fences driving a similar but non-significant trend. White was the only colour that influenced takeoff distances, with horses jumping over white fences having a larger takeoff distance. Overall, our results show that current obstacle coloration does not maximise contrast for horse vision, and that alternative colours may improve visibility and alter behavioural responses, with the ultimate goal of improving safety and welfare.

An Ishihara-style test of animal colour vision

Colour vision mediates ecologically relevant tasks for many animals, such as mate choice, foraging and predator avoidance. However, our understanding of animal colour perception is largely derived from human psychophysics, and behavioural tests of non-human animals are required to understand how colour signals are perceived. Here, we introduce a novel test of colour vision in animals inspired by the Ishihara colour charts, which are widely used to identify human colour deficiencies. In our method, distractor dots have a fixed chromaticity (hue and saturation) but vary in luminance. Animals can be trained to find single target dots that differ from distractor dots in chromaticity. We provide MATLAB code for creating these stimuli, which can be modified for use with different animals. We demonstrate the success of this method with triggerfish, Rhinecanthus aculeatus, which quickly learnt to select target dots that differed from distractor dots, and highlight behavioural parameters that can be measured, including success of finding the target dot, time to detection and error rate. We calculated discrimination thresholds by testing whether target colours that were of increasing colour distances (ΔS) from distractor dots could be detected, and calculated discrimination thresholds in different directions of colour space. At least for some colours, thresholds indicated better discrimination than expected from the receptor noise limited (RNL) model assuming 5% Weber fraction for the long-wavelength cone. This methodology could be used with other animals to address questions such as luminance thresholds, sensory bias, effects of sensory noise, colour categorization and saliency.

Prey and predators perceive orb-web spider conspicuousness differently: evaluating alternative hypotheses for color polymorphism evolution

Color polymorphisms have been traditionally attributed to apostatic selection. The perception of color depends on the visual system of the observer. Theoretical models predict that differently perceived degrees of conspicuousness by two predator and prey species may cause the evolution of polymorphisms in the presence of anti-apostatic and apostatic selection. The spider Gasteracantha cancriformis (Araneidae) possesses several conspicuous color morphs. In orb-web spiders, the prey attraction hypothesis states that conspicuous colors are prey lures that increase spider foraging success via flower mimicry. Therefore, polymorphism could be maintained if each morph attracted a different prey species (multiple prey hypothesis) and each spider mimicked a different flower color (flower mimicry hypothesis). Conspicuous colors could be a warning signal to predators because of the spider’s hard abdomen and spines. Multiple predators could perceive morphs differently and exert different degrees of selective pressures (multiple predator hypothesis). We explored these 3 hypotheses using reflectance data and color vision modeling to estimate the chromatic and achromatic contrast of G. cancriformis morphs as perceived by several potential prey and predator taxa. Our results revealed that individual taxa perceive the conspicuousness of morphs differently. Therefore, the multiple prey hypothesis and, in part, the multiple predator hypothesis may explain the evolution of color polymorphism in G. cancriformis, even in the presence of anti-apostatic selection. The flower mimicry hypothesis received support by color metrics, but not by color vision models. Other parameters not evaluated by color vision models could also affect the perception of morphs and influence morph survival and polymorphism stability.