Inverse resource allocation between vision and olfaction across the genus Drosophila

Divergent populations across different environments are exposed to critical sensory information related to locating a host or mate, as well as avoiding predators and pathogens. These sensory signals generate evolutionary changes in neuroanatomy and behavior; however, few studies have investigated patterns of neural architecture that occur between sensory systems, or that occur within large groups of closely-related organisms. Here we examine 62 species within the genus Drosophila and describe an inverse resource allocation between vision and olfaction, which we consistently observe at the periphery, within the brain, as well as during larval development. This sensory variation was noted across the entire genus and appears to represent repeated, independent evolutionary events, where one sensory modality is consistently selected for at the expense of the other. Moreover, we provide evidence of a developmental genetic constraint through the sharing of a single larval structure, the eye-antennal imaginal disc. In addition, we examine the ecological implications of visual or olfactory bias, including the potential impact on host-navigation and courtship.

surface area (µm 2 ) as compared to head size for each species. (G) Example of lateral and frontal views (Drosophila melanogaster), which were used to measure the body, head, eye and funiculus. (H) Plotting of the residuals, where neither body nor head size significantly correlate with the EF ratio trait, suggesting that this trait does not simply scale allometrically with respect to body and head size. (I) Residuals of head and body have highly similar deviations from EFratio, supporting that body and head size are highly correlated across all species. (J) Different chemosensory genes from 12-14 Drosophila species genomes and their correlation to the EF ratio 1 , where number of olfactory pseudogenes, for example, does not suggest a sensory tradeoff. (Data are provided at doi.org/10.17617/3.1D).   signifies vision or visual bias, while blue indicates olfaction or olfactory species. An asterisk denotes statistical significance between two groups (*P ≤ 0.05, ***P ≤ 0.001; T-test). (F) Sensillum counts from lambda scans from only the anterior side of the antenna and the comparisons between all six species. Means with the same letter are not significantly different from each other (ANOVA with Tukey-Kramer multiple comparison test). Error bars represent standard deviation. (G) There is no correlation between trichoid number and antennal surface area, arguing against the idea that larger species necessarily have more trichoids. (H) Absolute size comparisons between two species, illustrating the differences in body, head, and eye morphology, where the body of the D. suzukii female is 1.5 times larger, but possesses a 2.5 times larger eye than the D. melanogaster female. (Data are provided at doi.org/10.17617/3.1D). (E) Although each species differed in absolute size, the ratio of central brain to total or whole brain (OL, AL, and central brain) for each species was roughly the same.

Supplementary
(F) Schematic of measurements taken from different species. (G) Absolute size of components of the OL and the AL from each species. (H) Female and male wing pigmentation plotted against EF ratio, where there is a correlation between relatively larger eyes and wing pigment across both sexes. An asterisk denotes statistical significance between two groups (*P ≤ 0.05, ***P ≤ 0.001; T-test). (I) Data from courtship in light or dark conditions as tested against EF ratio, where there is a highly significant difference in EF ratio across the three groups of courtship. Here again, relatively larger eyes correlate with better performance in light conditions, or with complete light-dependence for courtship. Means with the same letter are not significantly different from each other (ANOVA with Tukey-Kramer multiple comparison test).

Supplementary Figure 4: Behavioral assays for visual and olfactory host navigation. (A)
Design of trap assays using several visual and olfactory objects in testing attractive stimuli for each species. Red was the most attractive against the white background for all species regardless of the odor type, and even without odor, red was sufficient to capture spotted wing species. There was no significant difference in attraction to red when in combination with the three tested odors. The only color difference between species was noted to be an attraction to green for D. suzukii, as well as blue when in combination with blueberries, which they were reared upon. (B) Petri dish behavioral assay comparing D. melanogaster and D. suzukii, where both species showed similar color preference when presented without odor, although when with an odor, D. suzukii had a higher tendency towards white, yellow, green, blue and red than the other species. (C) Reflection index and wavelength for each color used in the behavioral assays. (D) Two-choice trap assay, conducted in either full light, or full darkness. With lights off, all tested species were able to successfully navigate to the odor source; however, with lights on, the spotted wing species often mistakenly selected the visual object and not the odor object containing the fruit or food source, suggesting perhaps a visual bias or preference. In contrast, D. melanogaster always navigated to the odor source regardless of light condition or visual object, suggesting an olfactory bias or priority for this sensory cue. (Data are provided at doi.org/10.17617/3.1D).
Supplementary Figure 5: Antennal preparations and trichoid counts from selected species. (A) Each Drosophila species was mounted using single-sensillum recording (SSR) preparation techniques, and a series of images was taken to generate a z-stack photomontage. Trichoid sensilla were counted from male individuals over the same region of the funiculus for each Drosophila species. Images were taken with the arista mounted upward for consistency and for the best viewing angle as previously described for this sensillum type 2 . (B) Example of Drosophila species from a single phylogenetic clade that show a decreasing number of trichoid sensillum (left to right), and differences in surface area containing these sensilla, as well as differing sensillum length. developmental duration from egg to adult, we selected larvae for imaginal disc dissection during the same developmental window of time, namely the 3 rd instar wandering phase larvae, which occurs just prior to the onset of pupation. (B) Example of 3 rd instar larvae feeding on top layer of food (left) and 3 rd instar wandering phase larvae (right) that have stopped feeding and are in search of a suitable pupation site. The latter of which were selected from each species for consistent dissection of the imaginal disc. (Data are provided at http://doi.org/10.17617/3.1D)

Supplementary
Supplementary Table 1: All scientific names, rearing media and stock numbers. (A) Drosophila species in alphabetical order, in conjunction with media used for rearing, as well as stock center identity. More information about each species is available through these stock numbers (e.g. site of insect collection, collection date, and reference specimens) (B-C) Recipe for diets used in this study. Green and blue colored diets were supplemented with either Opuntia cactus powder or fresh blueberries to enhance oviposition. Flies were maintained in a density-controlled manner, with 20-25 females per vial.