The spectrum of light that an animal sees – from ultraviolet to far red light – is governed by the number and wavelength sensitivity of a family of retinal proteins called opsins. It has been hypothesized that the spectrum of light available in an environment influences the range of colors that a species has evolved to see. However, invertebrates and vertebrates use phylogenetically distinct opsins in their retinae, and it remains unclear whether these distinct opsins influence what animals see, or how they adapt to their light environments. Systematically utilizing published visual sensitivity data from across animal phyla, we found that terrestrial animals are more sensitive to shorter and longer wavelengths of light than aquatic animals, and that invertebrates are more sensitive to shorter wavelengths of light than vertebrates. Controlling for phylogeny removes the effects of habitat and lineage on visual sensitivity. Closed and open habitat terrestrial species have similar spectral sensitivities when comparing across the Metazoa, and deep water animals are more sensitive to shorter wavelengths of light than shallow water animals. Our results suggest that animals do adapt to their light environment, however the invertebrate-vertebrate evolutionary divergence has limited the degree to which animals can perform visual tuning.

Paper selection: We conducted Google Scholar searches in October 2017 and January 2018. Our first search used the search pattern “visual pigment” OR opsin OR “absorbance spectrum” “λ max” -human -man -men -woman -women -“Homo sapiens” -disease -regeneration. We conducted a second Google Scholar search using the search pattern visual pigment, opsin sensitivity, absorbance spectrum. For both searches, we excluded citations and patents.

We reviewed candidate articles using a three-step process. First, we screened by title and abstract to identify original research articles and review papers that examined animal visual physiology. We then screened articles to determine if they used microspectrophotometry, electrophysiology, pigment extraction, or in vitro mRNA expression followed by spectrophotometry, and that they measured visual sensitivity or visual pigment absorption from at least two animals. Finally, we only kept articles which used animals that were wild-caught or reared in full-spectrum light conditions, to avoid any effects of artificial lighting on visual sensitivity [56,57].

For review articles, we determined whether the authors had included measurements of the mean wavelength of peak sensitivity (λ_max) of some population in the article’s figures or tables. We downloaded the corresponding primary sources and filtered them using the process described above.

Visual pigment sensitivity data: We recorded the following data for each species of each paper that passed our filters: 1) mean wavelength of peak sensitivity (λ_max) for each visual pigment measured; 2) number of animals measured (n); 3) standard deviation of the mean λ_max (SD) (when available); and 4) where animals were caught (when available). We calculated sampling error for visual pigments when possible.

Habitat data: We used standardized data sources to classify each species by habitat. Sources included field guides [58–60], public databases (BugGuide, <bugguide.net>, Butterflies and Moths of North America, <butterfliesandmoths.org>, FishBase <fishbase.org>, SealifeBase <sealifebase.org>, IUCN Redlist <iucnredlist.org>) and online encyclopaedias including Animal Diversity Web (<animaldiversity.org>) and Encyclopedia of Life (<eol.org>). After first classifying species as terrestrial or aquatic, we then defined terrestrial sub-habitats: rainforest, forest, woodland, shrubland, grassland, and desert. We recategorized these habitats into three habitat types based on canopy density. Rainforest and temperate forest were designated as “closed” habitats. Woodland was considered to have “intermediate” canopy density  [25]. Shrubland, grassland, and desert were classified as "open" habitats.

Aquatic habitats included river, stream, pond, lake, coastal, estuarine, open-water marine, bottom-dwelling marine, abyssopelagic, abyssodemersal, bathypelagic, and bathydemersal habitats. We recategorized these habitats into two habitat types based on salinity. River, stream, pond, and lake habitats were considered "freshwater" habitats; while coastal, estuarine, open-water marine, and bottom-living marine habitats were "marine" habitats. Animals considered 'coastal' were those described as living in water along the coast, near shore, or in estuaries. We also recategorized these habitats into two habitat types based on whether light was abundant or not. Abyssopelagic, abyssodemersal, bathypelagic, and bathydemersal habitats receive little or no sunlight due to their depth in the water column and were considered "lightless" habitats. Species that were considered by our sources as deep-water species were also considered species that lived in "lightless" habitats. All other habitats were considered "lit" habitats.

Finally, we used FishBase, Sealifebase or field guides to identify the minimum and maximum depths for each species. We then used these data to calculate average depth per species (D_average = (D_max+D_min) *2^-1).