Many ways to make darker flies: Intra‐ and interspecific variation in Drosophila body pigmentation components

Abstract Body pigmentation is an evolutionarily diversified and ecologically relevant trait with substantial variation within and between species, and important roles in animal survival and reproduction. Insect pigmentation, in particular, provides some of the most compelling examples of adaptive evolution, including its ecological significance and genetic bases. Pigmentation includes multiple aspects of color and color pattern that may vary more or less independently, and can be under different selective pressures. We decompose Drosophila thorax and abdominal pigmentation, a valuable eco‐evo‐devo model, into distinct measurable traits related to color and color pattern. We investigate intra‐ and interspecific variation for those traits and assess its different sources. For each body part, we measured overall darkness, as well as four other pigmentation properties distinguishing between background color and color of the darker pattern elements that decorate each body part. By focusing on two standard D. melanogaster laboratory populations, we show that pigmentation components vary and covary in distinct manners depending on sex, genetic background, and temperature during development. Studying three natural populations of D. melanogaster along a latitudinal cline and five other Drosophila species, we then show that evolution of lighter or darker bodies can be achieved by changing distinct component traits. Our results paint a much more complex picture of body pigmentation variation than previous studies could uncover, including patterns of sexual dimorphism, thermal plasticity, and interspecific diversity. These findings underscore the value of detailed quantitative phenotyping and analysis of different sources of variation for a better understanding of phenotypic variation and diversification, and the ecological pressures and genetic mechanisms underlying them.

1995; Guillermo-Ferreira et al. 2014) or different species (e.g. predator avoidance via camouflage or aposematism; e.g. Reichstein et al. 1968;Futahashi and Fujiwara 2008;66 van Bergen and Beldade 2019), as well as thermoregulation (e.g. Rajpurohit et al. 2008;Sibilia et al. 2018). Moreover, insect pigmentation is tightly associated with various 68 other traits that are closely related to fitness (see Wittkopp and Beldade 2009;Mckinnon and Pierotti 2010). The diversity of insect pigmentation across species, 70 populations, sexes, and individuals of the same sex has been the focus of many eco-evodevo studies, providing key insight into the genetic basis of variation in pigmentation 72 (e.g. Pool and Aquadro 2007;Futahashi and Fujiwara 2008;Miyagi et al. 2015;Massey and Wittkopp 2016;Zhang et al. 2017;Orteu and Jiggins 2020) and exploring important 74 phenomena such as developmental plasticity (e.g. Solensky and Larkin 2009;Shearer et al. 2016;Monteiro et al. 2020), the origin of novelty (e.g. Shirai et al. 2012;Vargas-76 Lowman et al. 2019), and evolutionary constraints (Beldade et al. 2002b;Allen et al. 2008). 78 Variation in body pigmentation can arise from differences in color and/or in the 80 spatial arrangement of colors into specific patterns. These two aspects rely on largely distinct classes of genes involved in pigmentation development: those encoding the 82 enzymes responsible for pigment synthesis, and those encoding the transcription factors regulating those enzymes' expression at the appropriate time and place (see True 2003; 84 Wittkopp et al. 2003;Wittkopp and Beldade 2009). Changes in genes associated with each of these steps can result in changes in pigmentation between individuals and 86 between body parts (e.g. Wittkopp et al. 2002). In this respect, body pigmentation can be thought of as a multi-dimensional trait, made up of several components representing 88 aspects of actual color and of color pattern, and which might develop and evolve more or less independently. This has been explored in studies focusing on specific color 90 pattern elements, including on butterfly wings (e.g. Nijhout 2001;Monteiro 2015; Beldade and Peralta 2017), as well as on fly wings and abdomens (e.g. Jeong et al. 92 2006; Werner et al. 2010). Yet, rarely do studies of body pigmentation variation combine quantitative analysis of multiple color and color pattern traits.  Wittkopp et al. 2003;Gibert et al. 2007;Pool and Aquadro 2007;Massey and Wittkopp 2016). These studies characterized effects of environmental factors, such 100 as nutrition (e.g. Shakhmantsir et al. 2014) and temperature (e.g. David et al. 1990), as well as allelic variants of both subtle (e.g. Bastide et al. 2013) and large phenotypic 102 effect (e.g. Carbone et al. 2005). Variation in Drosophila pigmentation has been associated to clinal and seasonal variation in desiccation resistance, thermo-regulation, 104 and UV protection (e.g. Rajpurohit et al. 2008;Matute and Harris 2013;Parkash et al. 2014), and shown to correlate with other traits, such as reproductive success, behavior, 106 and immunity (e.g. Dombeck and Jaenike 2004;Takahashi 2013;Massey et al. 2019).
While studies of Drosophila pigmentation have included focus on different body parts 108 (e.g. trident on thorax, e.g. David et al. 1985; melanic patches on wings, e.g. True et al. 1999; dark bands of abdominal segments, e.g. Dembeck et al. 2015), these studies 110 typically analyze single and often qualitative properties of pigmentation (but see e.g. Saleh Ziabari and Shingleton 2017). Indeed, the detail in quantitative phenotyping of 112 body pigmentation does not match the sophistication of the analysis of its genetic and developmental bases. This is not unique to Drosophila pigmentation; the need for more 114 attention to be given to phenotyping has been called for repeatedly (Gerlai 2002;Houle et al. 2010;Kühl and Burghardt 2013;Deans et al. 2015;Laughlin and Messier 2015). 116 Here, we quantify various traits encompassing aspects of both color and color 118 pattern of abdomen and thorax pigmentation in Drosophila adults. We investigate how each of these pigmentation components (or traits) and the associations between them 120 differ between genotypes and developmental temperatures, within and across species.
We show that different pigmentation components can vary rather independently, and 122 that fly bodies can be made lighter or darker by changing different pigmentation components. We discuss our results in the context of the potential for evolutionary 124 diversification of pigmentation.

130
To investigate patterns and sources of variation in Drosophila body pigmentation, we developed a quantitative method to define five pigmentation traits that include aspects 132 of color and color pattern (see Figure S1 and Materials and Methods). We focused on the dorsal surface of thoraxes and abdomens, characterized for having different types of 134 dark "pattern elements" on a lighter "background" color: a trident at the center of the thorax and posterior bands on each segment of the abdomen. Flies were imaged under a 136 binocular scope in controlled light conditions. For each body part, we defined a transect between an anterior and a posterior landmark and collected color information for each 138 pixel along these transects ( Figure 1A, Figure S1). Using that information, we quantified a series of traits for each body part: overall darkness (Odk), relative length of 140 transect occupied by the darker "ornamental" pattern (Pat), actual color of both background (Cbk) and "ornamental" pattern elements (Cpa), and the distance in RGB 142 space between the darkest and the lightest that corresponds to the range of color variation (Ran). We investigated how these pigmentation components vary and co-vary 144 between sexes and between rearing temperatures in D. melanogaster representing standard laboratory strains, and natural populations from different geographical 146 locations, as well as in five additional Drosophila species. For each dataset (D. melanogaster laboratory strains, D. melanogaster clinal populations, and Drosophila 148 species), the multivariate multiple regression analyses showed that pigmentation differed significantly between strains/genotypes/species, sexes, and temperatures, with 150 effects that depended on body part (Table S1).

152
Variation in body pigmentation in D. melanogaster laboratory populations 154 We reared flies from two common laboratory genetic backgrounds (or strains) of D. melanogaster, Oregon R (OreR) and Canton S (CanS), at either 17°C or 28°C to assess 156 thermal plasticity and sexual dimorphism in our pigmentation traits (Figure 1, 2, Figure   S2, Table S2). We confirmed known patterns of thermal plasticity and sexual 158 dimorphism for body pigmentation, with flies reared at lower temperature being generally darker than those reared at higher temperature, and males being darker than 160 females ( Figure 1B, 2A, Figure S2). However, we found differences between strains and body parts in the extent, and sometimes the direction of both thermal plasticity and 162 sexual dimorphism for our pigmentation traits ( Figure 1B, 2A, Figure S2, Table S2), as well as for the correlations between them ( Figure 3A).

164
For overall darkness (Odk; dot plots in Figure 2A), flies reared at 17°C were 166 generally darker than those from 28°C, with the exception of CanS males (where differences were not significant in either body part), and OreR females (where 168 abdomens were darker in flies from 28°C). The abdomens were lighter in females relative to males (except for CanS from 17°C), but the thoraxes were lighter in males 170 relative to females (except for CanS from 28°C and OreR from 17°C). We also observed differences between sexes and temperatures for the other pigmentation traits 172 (Pat, Ran, Cbk, and Cpa; radar plots in Figure 2A; dot plots in Figure S2, Table S2), which depended on body part. While for the thorax the most striking differences were 174 seen in Ran (for females between temperatures), for the abdomen they were seen for Pat (distinguishing females from 28°C from others) and Ran (extreme for OreR females) 176 ( Figure S2). Variation was only loosely correlated between traits, with few significant correlations, which differed between genetic backgrounds, sexes, and rearing 178 temperatures. Overall, correlations between traits were weaker across body parts relative to within body parts, and in males relative to females ( Figure 3A).

180
For those pigmentation traits found to be thermally plastic (i.e. significant 182 differences between individuals reared at different temperatures; cf. Figure S2, Table   S2), we investigated which stages of development were thermally responsive. To do so,   Figure 3B), equal to one of the 194 extreme temperatures (Pat; Figure 3B), or more extreme than both T17 and T28 (Odk; Figure 3B). The period when exposure to a different temperature significantly affected 196 phenotype also differed between traits and genetic backgrounds ( Figure 3B, Figure S3).

Body pigmentation differences between D. melanogaster natural populations and
Drosophila species 200 We quantified variation in pigmentation components in wild-caught populations  Table S4). Geographical populations differed in 208 overall darkness (Odk; dot plots in Figure 2B) and in color (both Cbk and Cpa) for the abdomen, but not the thorax ( Figure 2B, Figure S5, Table S4). For the thorax, only Ran 210 and Pat differed between locations ( Figure 2B, Table S4). Most pigmentation traits (except thorax color; Cpa and Cbk) were thermally plastic, with darker flies for 212 development at 17°C relative to 28°C ( Figure 1C, 2B, Figure S4). The Northern-and Southern-most populations (i.e. Finland and Spain, respectively) did not necessarily 214 show the most extreme phenotypes, neither in terms of overall darkness nor in the extent of plasticity therein ( Figure 2B, Figure S5). We also found significant differences 216 between isofemale genotypes (and their plasticity) within each geographical location ( Figure 2B, Table S4).  Figure 1D, 2C). We found differences between species in extent and direction of sexual dimorphism and of 224 thermal plasticity for the different pigmentation traits ( Figure 1D, 2C, Figure S2B, Table S5). For instance, for Odk (dot plots in Figure 2C), while D. malerkotliana 226 showed no differences between temperatures and clear differences between sexes, D.
simulans had very high thermal plasticity but reduced sexual dimorphism (no 228 differences between females and males reared at 17°C). For the other pigmentation traits (radar plots in Figure 2C and dot plots in Figure S2B), larger differences between 230 sexes and/or temperatures were observed for Pat and/or Ran, and less for actual colors (Cpa and Cbk).

234
We decomposed Drosophila body pigmentation into different quantitative traits, 236 including overall darkness (Odk), and traits reflecting properties of color and color pattern (Pat, Ran, Cbk, and Cpa) of both thoraxes and abdomens. We showed  David et al. 2002). In D. melanogaster for instance, most work has focused on abdominal pigmentation, and specifically on the dark bands of the 256 posterior-most segments, which is sexually dimorphic (males are generally darker than females; e.g. Kopp et al. 2000) and thermally plastic (flies from lower developmental 258 temperatures are generally darker than flies from higher developmental temperatures; e.g. David et al. 1990;Gibert et al. 2007Gibert et al. , 2009. We extended the analysis of body 260 pigmentation to quantifying different properties of both abdomen and thorax pigmentation in D. melanogaster and other Drosophila species. This more detailed 262 analysis ultimately painted a more complex picture of variation in Drosophila body pigmentation. We did not, for instance, always find that males were darker than 264 females, or that flies reared at lower temperatures were darker than those from higher temperatures, but rather, we found trait specificities in how pigmentation varied 266 between sexes and between developmental temperatures. This was true for overall darkness (Odk) of the abdomen, the trait that would presumably be more similar to 268 previous (largely qualitative) characterizations of abdominal pigmentation (e.g. David et al. 1990;Hollocher et al. 2000a), but also for other properties of body pigmentation, 270 including actual color of background and pattern elements (i.e. abdominal bands and thoracic trident). Moreover, we also showed that pigmentation components, as well as 272 sexual dimorphism and thermal plasticity therein, vary greatly between species, genotypes, and body parts. The mechanisms underlying such intra-and inter-specific  Studies exploring the ecological conditions driving the evolution of melanism in Drosophila have documented correlations between body pigmentation and several eco-300 geographic variables (e.g. latitude, altitude, temperature, humidity) (e.g. Gibert et al. 2016;Shearer et al. 2016). Clinal variation in pigmentation, for 302 instance, has been shown for thoracic trident (e.g. David et al. 1985;Telonis-Scott et al. 2011) and for abdominal pigmentation (e.g. Pool and Aquadro 2007;Das 2009). 304 Generally, darker phenotypes in colder environments (e.g. at high latitudes or altitudes) have been hypothesized to allow flies to better absorb solar radiation (c.f. thermal 306 budget or thermal melanism hypothesis; Trullas et al. 2007;Clusella-Trullas et al. 2008), to increase desiccation resistance (e.g. Parkash et al. 2008), and/or to provide

372
For the experiment of the windows of sensitivity for pigmentation, we exposed developing flies to 17°C or 28°C during one window of development while they were 374 kept at 23°C for the remaining stages. We tested four different treatments at 17°C and at     472 We used post-hoc pairwise comparisons (Tukey's honest significant differences) to identify differences between strains, sexes, temperatures and/or thermal regimes. Facility at the IGC for support in setting up the image acquisition system. 502 We declare that no conflict of interest exists. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.     Positive correlations are shown in purple and negative correlations in orange.

824
Correlations not statistically significantly different from zero (p-value > 0.05) are indicated with a cross. B. Pigmentation traits (Y axis) in females of two D. 826 melanogaster genetic backgrounds (CanS and OreR) exposed to each of the temperature regimes during development (X axis). The thermal regimes codes and corresponding 828 stages that were exposed to either 17°C or 28°C (instead of the basal temperature of 23°C) were: T (constant temperature), L (late larval development), p (early pupal 830 period) and, P (late pupal period). In each graph, dots represent phenotypes of single individual females, and the horizontal bar is the mean of those values. The results of the 832 test for differences between strains and thermal regimes on the different plastic traits are shown in Table S3. Letters indicate results of post-hoc pairwise comparisons between 834 groups: different letters when significantly different (p-value<0.05 for Tukey's honest significance test).      T17  L17  p17  P17  T23  P28  p28  L28  T28  T17  L17  p17  P17  T23  P28  p28  L28