Melanic pigmentation and light preference within and between two Drosophila species

Abstract Environmental adaptation and species divergence often involve suites of co‐evolving traits. Pigmentation in insects presents a variable, adaptive, and well‐characterized class of phenotypes for which correlations with multiple other traits have been demonstrated. In Drosophila, the pigmentation genes ebony and tan have pleiotropic effects on flies' response to light, creating the potential for correlated evolution of pigmentation and vision. Here, we investigate differences in light preference within and between two sister species, Drosophila americana and D. novamexicana, which differ in pigmentation in part because of evolution at ebony and tan and occupy environments that differ in many variables including solar radiation. We hypothesized that lighter pigmentation would be correlated with a greater preference for environmental light and tested this hypothesis using a habitat choice experiment. In a first set of experiments, using males of D. novamexicana line N14 and D. americana line A00, the light‐bodied D. novamexicana was found slightly but significantly more often than D. americana in the light habitat. A second experiment, which included additional lines and females as well as males, failed to find any significant difference between D. novamexicana‐N14 and D. americana‐A00. Additionally, the other dark line of D. americana (A04) was found in the light habitat more often than the light‐bodied D. novamexicana‐N14, in contrast to our predictions. However, the lightest line of D. americana, A01, was found substantially and significantly more often in the light habitat than the two darker lines of D. americana, thus providing partial support for our hypothesis. Finally, across all four lines, females were found more often in the light habitat than their more darkly pigmented male counterparts. Additional replication is needed to corroborate these findings and evaluate conflicting results, with the consistent effect of sex within and between species providing an especially intriguing avenue for further research.


| INTRODUC TI ON
Correlations among phenotypic traits are ubiquitous, with profound implications for the evolution of populations (Lande and Arnold, 1983). Although phenotypic correlations are frequently observed in nature, the underlying causes are potentially numerous and are often unknown (Endler, 1986;Stearns, 1992). Traits can be genetically correlated due to either linkage or pleiotropy, while genetically unassociated traits may evolve in a correlated fashion due to "selective covariance," in which selection tends to act simultaneously on two or more traits (Armbruster & Schwaegerle, 1996). Finally, populations and species can diverge from one another in suites of traits due simply to the unique history of mutation, migration, and drift within each group (Armbruster & Schwaegerle, 1996).
One trait that frequently evolves as part of a suite of correlated characters is pigmentation. In the model insect genus Drosophila, correlations due to pleiotropy of an underlying gene have been reported for pigmentation and trichome patterns (Gompel & Carroll, 2003) and for pigmentation and vision (True et al., 2005). Selective covariance is also likely to influence patterns of pigment evolution in Drosophila: altitudinal and latitudinal gradients in melanic pigmentation have been documented in multiple species and have been ascribed to selection associated with heat, ultraviolet radiation, and/or humidity (Clusella Trullas et al., 2007;Pool & Aquadro, 2007;Rajpurohit & Nedved, 2013;Rajpurohit et al., 2008;Telonis-Scott et al., 2011;True, 2003). Thus, pigmentation in Drosophila is a promising system for investigating both genetic and environmental influences on the evolution of correlated traits.

While most of the documented pigmentation clines in
Drosophila are altitudinal or latitudinal, a unique longitudinal gradient has been observed in Drosophila americana, with very dark brown flies found in the eastern United States and lighter flies found as far west as the Rocky Mountains (Throckmorton, 1982).
Sister species D. novamexicana features an evolutionarily derived, lighter, and yellower body color, and its geographical distribution in the desert Southwest of the United States makes it appear to be a geographic extension of the pigmentation cline in D. americana . Pigmentation in D. novamexicana is also highly variable, but it is always lighter than even the lightest lines of D. americana (Davis & Moyle, 2019). In addition to these patterns of variation within and between species (Figure 1a), female D. americana have been shown to be slightly lighter in color compared to consistent effect of sex within and between species providing an especially intriguing avenue for further research. The balance of ebony and tan expression helps determine cuticular pigmentation. (c) The same genes, ebony and tan, also participate in histamine recycling in the visual system. (b) and (c) are redrawn from Takahashi (2013) males of the same lines despite a lack of difference in color patterning (Wittkopp et al., 2011).
americana line DN12 and D. novamexicana line N14, and ebony and tan have been shown to be the causal genes within these QTLs (Lamb et al., 2020;. The Ebony and Tan enzymes catalyze reverse reactions in the melanin/sclerotin pigment biosynthesis pathway (Figure 1b), with Ebony promoting the synthesis of yellow sclerotin pigment and Tan promoting the synthesis of brown and black melanin (Wittkopp & Beldade, 2009).
Pigmentation trends both within and between these two species covary with environmental factors across the United States. The range of the light-bodied D. novamexicana is characterized by higher temperatures, more solar radiation, and less moisture compared to the range of D. americana (Davis & Moyle, 2019). Consistent with its desert environment, D. novamexicana is significantly more tolerant of desiccation than D. americana (Davis & Moyle, 2020). Within D. americana, the adaptive cline reported by Wittkopp et al. (2011) showed no association between pigment variation and altitude, mean temperature, or relative humidity, and a manipulative experiment ruled out direct effects of pigmentation on desiccation tolerance. A re-analysis of that dataset by Clusella-Trullas and Terblanche (2011), with additional variables, provided support for an association between pigmentation, light, and temperature range: the darker D.
americana populations, found in the eastern United States, tend to be in locations with lower mean solar radiation and lower diurnal temperature ranges.
The connection between pigment and environmental light is particularly intriguing, because the pigmentation genes ebony and tan both have pleiotropic effects on fly responses to light (Takahashi, 2013;Figure 1b,c). The Tan enzyme is produced not only in developing cuticles but also in the photoreceptors of the eye, where it processes the inactive compound carcinine (also known as N-beta-alanyl histamine) into the neurotransmitter histamine. When a light signal is received, histamine is released by photoreceptors into the synaptic cleft to propagate the signal; from there, it is removed to the associated glial cells, where Ebony converts it back to carcinine, to be returned once more to the photoreceptors (Gavin et al., 2007).
The dark-colored ebony mutants of D. melanogaster show reduced mating success relative to wild-type flies under regular laboratory conditions, but higher mating success than wild-type flies in dim light (Kyriacou, 1981;Kyriacou et al., 1978;Rendel, 1951) The same alleles of ebony and tan that confer lighter, yellower coloration in D. novamexicana are also found in some though not all light-colored populations of D. americana, indicating that the genetic basis for light body color is partially shared within and between species (Sramkoski et al., 2020; 3. between females and males of the same lines.
Based on melanic coloration, we predicted higher light pref- Our data provide preliminary evidence that pigmentation may be correlated with light-seeking behavior in the D. americana-D. novamexicana species pair.

| Fly lines
Drosophila americana lines A04, A00, and A01 and Drosophila novamexicana line N14 were ordered from the Cornell University Drosophila Stock Center (Table 1) and maintained at Whitman College on Nutri-Fly Instant fly food (Genesee Scientific, San Diego, CA, USA). Flies were maintained at ambient light, on benches adjacent to windows.
Within D. americana, A01 is the lightest line that has been documented to date, and it contains a novamexicana-like (functionally "light") allele linked to the tan gene, while the dark A00 line contains functionally "dark" alleles at both ebony and tan ). The dark A04 line is functionally uncharacterized, although it is phenotypically very similar to line A00 (Table 1).
Drosophila novamexicana-N14 is the best characterized line of its species (Cooley et al., 2012;  pandemic, data collection by two of the experimenters was split between work done at Whitman College and work done at the students' homes. In each case, the data were coded as two separate experiments based on their locations.

| Cage construction
To provide alternate light environments for the behavioral choice experiments, cages were constructed using small, transparent betta fish tanks with a dark plastic divider ( Figure 2b). All outer sides of half of each cage were covered in two layers of duct tape to create a dark environment. Uniform holes ¼" in diameter were drilled into the dividers, allowing flies to pass between the light and dark sides of the cages. The dividers were locked in place by hot glue, sealing them to the insides of the cages. Clear tape was used on the inside of the lids to prevent flies from escaping through airholes. Each side of the container had identical plastic caps filled with synthetic fly food to sustain the flies throughout the trial period. Only enough water was added to the synthetic fly food to slightly saturate it, to prevent the buildup of excess condensation in the cages.

| Selection of flies for behavioral trials
To ensure that flies used in the behavioral trials were no more than 1 week old, all adult flies were transferred out of the collecting vials 1 week prior to each trial. On the day of the trial, the collecting vials-containing flies which had eclosed within the past weekwere chilled at 4℃ to immobilize the flies. Genital morphology was used to sex the flies, since these species lack both sex combs and sex-specific pigmentation patterns. Flies of a single sex and taxon were sorted in sets of five into empty test tubes. The vials were kept off ice so liveliness could be evaluated once they warmed up.
This was to ensure they had not been damaged and could fly and move normally. Flies that appeared old, deformed, or injured were also returned to the main population. Once collected and checked for liveliness, flies were re-immobilized by chilling on ice to facilitate transfer and were then poured into each side of the cage. The lids were secured with clear tape.

| Data collection in the behavioral trials
In 2017, fly cages were placed in a darkened room under a greenhouse grow light set on a 12-hr timer. Due to concerns that the was added.
At the end of each trial, cages were placed in a freezer at −20℃ for 1 hr to immobilize the flies. This allowed us to remove the lid and more thoroughly look for missing or dead flies. The data from cages with dead or missing flies were excluded from analysis. We disposed of the flies and cleaned the cages with ethanol.

| Temperature evaluation
In the 2019 experiment, a temperature control study was set up to test for a temperature difference between the light and dark sides of the cages. The wire probes of Fluke 52 II dual input digital thermometers (Everett, WA) were placed in both the light and dark sides of two empty cages. We recorded the temperature reading of each side of each cage, at noon and 4 p.m. daily for 6 days.

| Statistical analyses
To test for differences in fly light preference, a generalized linear model was fitted using the glm() command in RStudio 1.3.1093, "Apricot Nasturtium," within the lme4 package. We assumed a Poisson distribution for the dependent variable, which was the num- A paired t test in R was used to determine whether there was a significant temperature difference between the light and dark sides of the cages.

| Genotyping
At  (Table 2). This result is consistent with our hypothesis that the light-bodied D. novamexicana, which is found in putatively lighter and brighter habitats in the wild, would show a stronger preference for well-lit environments than the dark-bodied

D. americana.
The behavioral difference between species cannot be ascribed to a preference for distinct temperature regimes: the mean difference in temperature between the light and dark habitats was negligible, at both noon and 4 p.m., and not statistically significant ( Figure 4; t = 0.848, df = 3, p = .405). Time of day had a significant effect on total numbers of flies in the light habitat ( Table 2). Flies of both species were found in the light habitat more often at 4 p.m.
than at 12 p.m. (Figure 3). Thus, time of day affected the total numbers of flies on the light side, but did not alter the observed pattern of greater light preference in D. novamexicana compared to D. americana.

| Single-taxon trials of males and females show varied effects of taxon and a consistent effect of sex
In experiments with one taxon per cage, in contrast to the mixedspecies experiments, no significant difference was observed between D. americana-A00 and D. novamexicana-N14 (Table 2).
The preference of D. novamexicana for the light habitat was similar to that of the dark-bodied A04 and A00 lines of D. americana ( Figure 5). Within D. americana, the lightest line (A01) was found in the light habitat more often than either of the darker lines (A00, A04).
In the 2020 experiments, a consistent and significant effect of sex was observed (Table 2). Across all four lines of flies utilized, females-which have slightly lighter abdominal pigmentation than males-were observed more often than males in the light habitat ( Figure 5).  Note: Data were collected from each cage once per day for 6 days. Taxon and sex were considered fixed effects; experiment and cage were considered random effects; and the response variable (the number of flies in the "light" habitat each day) was assumed to have a Poisson distribution. A positive Z-value indicates a greater number of flies in the "light" habitat relative to A00 (for effects of taxon); females (for effect of sex); or the 12 p.m. time point (for effect of time of day). N = the number of successful 6-day trials across both sexes and all taxa, with success based on all flies being present and alive at the end of the 6 days. ns, not significant (p > .05).

F I G U R E 4
Cage temperature is consistent across habitats. Bars represent the range, boxes represent quartiles, and horizontal lines inside the boxes mark the median. Sample size is shown above each data column. Data were collected once per day, for 6 days, on each of two cages, in 2019. Temperature did not differ significantly between light habitat and dark habitat (paired t test: t = 0.848, df = 23, p = .405) Seasonal variation might be expected to influence fly behavior, especially given that seasonality in Drosophila appears to depend on a circadian clock (Stoleru et al., 2007) which in turn is influenced by ebony (Newby & Jackson, 1991;Suh & Jackson, 2007). While we cannot exclude the effects of seasonality, we note that both of our sets of experiments included fall, spring, and summer data collection efforts.
Alternatively  (Caletka & McAllister, 2004). Additionally, we found that the mixed-species trials produced a greater species difference in habitat choice compared to single-taxon trials. In contrast, Gupta et al. (2019) found that aggressive behavior tended to be lower toward heterospecifics than toward conspecifics, which would if anything tend to promote coexistence rather than spatial segregation of the two species.
In a comparison of courtship and mating behaviors in D. americana and D. novamexicana, Spieth (1951) found that D. novamexicana males were more active and aggressive in pursuing mating attempts than D. americana males. This could lead to interspecific dynamics impacting the results of the 2017, 2018, and 2019 datasets, although male-male interactions per se were not addressed in that study (Spieth, 1951). Given the relatively small effect of species, and the variation observed across experiments, additional research will be required to determine the robustness and replicability of the species difference documented here.
In our second set of experiments, we explored the effects of intraspecies pigment variation and sex on habitat choice. Pigment variation within D. americana was somewhat correlated with habitat choice: the lightest line (A01) was found significantly more often in the light habitat than the two dark lines (A04 and A00). Line A01 has a functionally D. novamexicana-like ("light") allele at tan, but not ebony, whereas line A00 has non-novamexicana-like ("dark") alleles at both genomic regions . Given the pleiotropic role of tan in recycling histamines in the visual system, it is possible that the A01 "light" allele at the tan locus contributes to that line's apparently greater preference for well-lit habitats. Across D. americana, the genetic basis of pigment variation is complex and is only incompletely explained by variation at tan and ebony (Sramkoski et al., 2020). Future research on the potential pleiotropic effects of tan and ebony is thus best done on fly lines such as A01 and A00, whose tan and ebony alleles have been functionally characterized . Because the genetic basis for pigmentation in the dark line A04 is unknown, and tan and ebony might not be major contributors, we consider predictions regarding line A04 to be less robust than predictions regarding lines A01 or A00.
Interestingly, our second set of experiments also revealed a significant effect of sex. Female flies were found in the light habitat more often than males, in D. novamexicana as well as in all three lines of D. americana. Within D. americana, females have slightly lighter melanin pigmentation than males (Wittkopp et al., 2011). This finding is, therefore, consistent with our hypothesis that lighter bodied flies will have a correlated preference for lighter habitats. Although many sex-linked behaviors have been reported in Drosophila (Asahina, 2018), sex-specific differences in light preference have not, to our knowledge, been previously demonstrated.
Overall, our findings in D. americana and D. novamexicana suggest that correlations may exist between pigmentation and habitat choice between species, within species, and between the sexes, with trends in each case for lighter pigmentation to be associated with a slightly greater preference for a brightly lit environment. Out of seven comparisons made, four support a positive correlation between light body color and light habitat preference; two support a negative correlation; and one supports no correlation (Table 3).  The work presented here is one of few behavioral studies of these two species (but see Spieth, 1951) and the first demonstration to our knowledge of a sex-specific difference in preference for environmental light in Drosophila. Given the variation of our findings for D. novamexicana between our two experimental designs, additional replication will be necessary to evaluate the correlations that we observed between pigmentation and behavior. However, the majority of our comparisons suggest a pattern in which lighter bodied flies tend to exhibit preference for a more brightly lit environment. Two genes, tan and ebony, together explain most of the color difference between the dark-bodied D. americana-DN12 and the lighter bodied D. novamexicana-N14 (Lamb et al., 2020; and are also required for visual function (Heisenberg, 1972;Takahashi, 2013;True et al., 2005). We propose that the pleiotropic nature of tan and ebony may have shaped evolutionary change in both pigmentation and light preferencepotentially within as well as between these two closely related yet intriguingly divergent species.

ACK N OWLED G M ENTS
We thank Abigail Lamb and Patricia J. Wittkopp for a generous donation of fly lines used in pilot studies; Emily Hamada for assistance in managing the laboratories; and Patricia J. Wittkopp and two reviewers for helpful comments on the manuscript. SR and IW were supported by Whitman College faculty-student research awards; CEV and SML were supported by Whitman College Abshire awards; AMC was supported by NSF-DEB 1655311 and NSF-DEB 1754075.

CO N FLI C T O F I NTE R E S T
None declared.

O PEN R E S E A RCH BA D G E S
This article has earned an Open Data Badge for making publicly available the digitally-shareable data necessary to reproduce the reported results. The data are available at https://doi.org/10.5061/ dryad.dv41n s1xz.