Pupal size as a proxy for fat content in laboratory-reared and field-collected Drosophila species

In arthropods, larger individuals tend to have more fat reserves, but data for many taxa are still missing. For the vinegar fly Drosophila melanogaster, only few studies have provided experimental data linking body size to fat content. This is rather surprising considering the widespread use of D. melanogaster as a model system in biology. Here, we hypothesized that fat content in D. melanogaster is positively correlated with body size. To test this, we manipulated the developmental environment of D. melanogaster by decreasing food availability. We then measured pupal size and quantified fat content of laboratory-reared D. melanogaster. We subsequently measured pupal size and fat content of several field-caught Drosophila species. Starvation, crowding, and reduced nutrient content led to smaller laboratory-reared pupae that contained less fat. Pupal size was indeed found to be positively correlated with fat content. The same correlation was found for field-caught Drosophila pupae belonging to different species. As fat reserves are often strongly linked to fitness in insects, further knowledge on the relationship between body size and fat content can provide important information for studies on insect ecology and physiology.

Pupal size as a proxy for fat content in laboratory-reared and field-collected Drosophila species Thomas Enriquez 1* , Victoria Lievens 2 , Caroline M. Nieberding 2 & Bertanne Visser 1 In arthropods, larger individuals tend to have more fat reserves, but data for many taxa are still missing. For the vinegar fly Drosophila melanogaster, only few studies have provided experimental data linking body size to fat content. This is rather surprising considering the widespread use of D. melanogaster as a model system in biology. Here, we hypothesized that fat content in D. melanogaster is positively correlated with body size. To test this, we manipulated the developmental environment of D. melanogaster by decreasing food availability. We then measured pupal size and quantified fat content of laboratory-reared D. melanogaster. We subsequently measured pupal size and fat content of several field-caught Drosophila species. Starvation, crowding, and reduced nutrient content led to smaller laboratory-reared pupae that contained less fat. Pupal size was indeed found to be positively correlated with fat content. The same correlation was found for field-caught Drosophila pupae belonging to different species. As fat reserves are often strongly linked to fitness in insects, further knowledge on the relationship between body size and fat content can provide important information for studies on insect ecology and physiology.
Body size is a key life history trait in insects [1][2][3][4] . Many experimental studies consequently use body size measurements as a proxy for fitness (e.g. [5][6][7][8], because a larger size generally leads to a higher fecundity 9,10 and longevity 6,9 . Insect body size depends on the environmental conditions experienced during development, such as temperature 11 , nutrition 12,13 or population density 14 . As insects do not grow as adults, body size is determined entirely during the juvenile stages 15 . Final body size is thus reached after pupation or after the final molt for insects with complete or incomplete metamorphosis, respectively [16][17][18][19] . During the development of holometabolous insects, developing larvae need to reach several weight limits for metamorphosis to occur, i.e., the minimal viable weight and the critical weight 20,21 . The minimal viable weight corresponds to the weight at which nutrient reserves, such as fat, are sufficient to survive after metamorphosis, while the critical weight corresponds to a threshold after which metamorphosis can no longer be delayed, even if the larva is starved 20 . In Drosophila melanogaster, the minimal viable weight and the critical weight are occurring almost simultaneously during larval development 20 . Starvation before reaching the minimal weight impedes the initiation of metamorphosis, leading to death of the larva, while reduced food intake or starvation after reaching the critical weight leads to smaller adult individuals 20,22 . Reduced nutrition during development thus generally decreases size. A range of traits are correlated to body size (i.e., allometry), including the relative dimensions of body parts, as well as physiological and behavioral traits 23 . In arthropods, one such key trait is the amount of fat reserves used for energy storage 19,[24][25][26][27][28][29][30] . Lipids are essential macronutrients for nearly all living organisms, and most animals have the capacity to synthesize and store lipids when resources are abundant 31 . In insects, lipids are stored as lipid droplets in adipocytes of the fat body, mostly as triglycerides, i.e., fat 3132 . The fat body is an organ present in arthropods that consists of loose tissues mostly located in the abdomen. The fat body is involved in numerous metabolic functions and plays a key role in storage and release of lipids, with a function similar to the liver in vertebrates 31 . In insects, most lipids are accumulated during the larval stage and fat content generally peaks right before pupation 33,34 . Part of the fat reserves are then used for metamorphosis (~ 35% in D. melanogaster 34 ; up to 50% in the fruit fly Ceratitis capitata 35  www.nature.com/scientificreports/ on the amount of fat accumulated during the larval stage and the amount consumed during metamorphosis 33,34 . For adult insects, fat reserves can have a large effect on fitness, as high fat reserves can positively affect key life history traits, such as longevity 36 and fecundity 24,25 . Lipid reserves further play a key role for many other functions, such as stress resistance, survival during overwintering 37 , resistance to drought and starvation 36,38,39 , and increased immunity 40 . Several studies have investigated the link between fat content and size in Drosophila melanogaster. For instance, Bryk et al. 41 investigated the role of a protein (MAP4K3) on the regulation of body size in D. melanogaster, showing that flies where MAP4K3 was knocked down were smaller and had a lower fat content than controls. In another study, Gasser et al. 42 created lines selected for high mortality and compared body size and fat content with lines selected for low mortality conditions. Only slight differences were observed, but generally the high-mortality selected flies were smaller and leaner than flies from the control line. Chippindale et al. 43 also compared artificially selected lines, with unselected flies being smaller and containing less fat compared to individuals selected for starvation resistance. Kristensen et al. 44 selected flies for 17 generation on a high protein diet and observed that these individuals had a greater body size and contained more fat than their counterparts selected on a standard diet. Juarez-Carreño et al. 45 studied the role of candidate genes in body-fat sensing, and showed that larvae where the gene Sema1a (a gene regulating lipid transport and ribosome maturation) was knocked down were bigger and contained more fat than control larvae. These mutants were, however, unable to pupate and initiate metamorphosis. All these studies indeed report that smaller individuals have lower fat reserves (or conversely that bigger individuals have higher fat reserves), but for each of these studies treatments were compared to a control and variation in fat content was not directly correlated to size. Overall, relatively few studies have determined the relationship between body size and fat content in adult D. melanogaster, and immature developmental stages are rarely studied. This lack of knowledge is surprising given that D. melanogaster is a widely used model species in biology [46][47][48][49] , and a promising emerging model for studying lipid metabolism and obesity [50][51][52][53][54][55] .
To the best of our knowledge, no data yet exists on the relationship between size and fat content of non-mutant Drosophila prior to emergence. Here, we aimed to test whether fat reserves and size are positively correlated in Drosophila pupae. We further aimed to establish a non-invasive method to estimate pupal fat content (i.e., without destructive sampling) based on size. By manipulating developmental conditions, experienced by D. melanogaster larvae reared under laboratory conditions, in terms of nutrient content and availability, we produced a gradient of pupal sizes that were subsequently measured for total fat content. To be able to expand our findings to more ecologically relevant conditions, we further collected wild Drosophila species and tested for a correlation between pupal size and fat content. Our results show that there is a strong positive correlation between pupal size and fat content, both in laboratory-reared and field-caught Drosophila. Estimates of pupal size thus provide a good proxy for pupal fat content in several Drosophila species.

Methods
Insect maintenance and developmental conditions. Our Drosophila melanogaster (Diptera: Drosophilidae) stock originated from a culture that was set up in 1994 from field collections in Sainte-Foy-les-Lyon (France), kindly provided by Patricia Gibert (Claude Bernard University, Lyon, France) in 2016. Larvae were maintained in flasks with continuous access to food medium (60 ml/flask; composition: 20 g agar, 35 g yeast, 50 g sugar, 5 ml nipagin containing 100 g 4-methyl hydroxyl benzoate in 1 l 96% alcohol, and 5 ml propionic acid per liter water). After emergence, adults were maintained in cages (50 × 50 × 50 cm) with continuous access to the same food medium that was replaced every 3 to 4 days. Individuals were kept at a temperature of 23 °C, a relative humidity of 75%, and a photoperiod of L:D 16:8, unless stated otherwise.
To generate pupae that varied in size, we manipulated nutrient content or availability during fly development using three methods: starvation, crowding, and modification of the sugar content in the medium. To do so, flies were allowed to lay eggs during 24 h. Approximately 100 eggs were then collected using a fine paintbrush and distributed in vials (containing 10 ml food medium, a surplus quantity of food to avoid competition between larvae) that differed in nutrient content. To change the nutrient content of the food medium, the standard medium (described above; denoted as 1/1 = 1 part sugar/1 part yeast) was modified to contain either twice more (i.e., 2/1; n = 3 vials) or no sugar (i.e., 0/1; n = 1 vial). As a control, 4 vials of the standard 1/1 medium were also prepared. Nutrient availability was altered by allowing 100 larvae per vial to feed on the standard 1/1 medium for 2 or 3 days. Larvae were then starved by transferring them to a new vial containing a medium without sugar or yeast (starvation after 2d and 3d, respectively; n = 3 vials for both treatments). A third treatment was added where the number of individuals per vial was increased, leading to crowding, and hence a reduction in nutrient availability. To create crowding conditions, 300 eggs were counted under a stereomicroscope and transferred to a vial containing 1 ml of the standard 1/1 medium. This egg density (300 eggs ml −1 ) was chosen, because previous work showed that D. melanogaster experiences strong crowding under these conditions 56 . All vials were inspected twice a day for newly formed pupae to ensure that pupae were collected within one day after pupation. Pupae were collected individually, their development time recorded, and then frozen at − 20 °C until further processing. Only rarely vials could not be inspected within 24 h, which was taken into account in the statistical analyses (see below).
Collection of Drosophila pupae from the field. In addition to manipulating the size of laboratoryreared D. melanogaster, we aimed to investigate the correlation between size and fat content of wild Drosophila individuals. To do so, we collected individuals from the field using banana-bait traps. Each trap consisted of a 0.75 l plastic box, with an opening in the lid. The opening was covered by a fine net mesh of ~ 1 mm. Three traps were prepared, each containing half a banana, as well as a mixture of live baker's yeast and apple cider vinegar. www.nature.com/scientificreports/ Each trap was then attached to a tree with the opening facing downward for 1 week in a backyard in Leuven (Belgium). Temperature and humidity were monitored with a thermometer-hygrometer Ibutton (Maxim Integrated) placed inside one of the traps ( Supplementary Fig. 1a). For each trap, bananas containing eggs and larvae were placed in a flask containing the standard 1/1 food medium (three flasks in total). In addition to bananabait traps, we also collected cherries infested by Drosophila suzukii from a cherry tree at the same location in Leuven. Cherries were distributed among 6 flasks containing the standard 1/1 food medium. All flasks were then kept in a cage and placed outside our facility (facing North and in continuous shadow). Temperature and humidity inside the cage were also monitored using an Ibutton device ( Supplementary Fig. 1b). Pupae were subsequently collected as described above for laboratory-reared flies. Visual species identification based on pupae is very difficult for most drosophilid species, but the pupal shape of D. suzukii is easily recognizable 57 . Drosophila obtained from cherries were, therefore, identified as D. suzukii based on the shape of their respiratory tubes. Pupae obtained from banana-bait traps remained largely unidentified (referenced hereafter as "other species").
To have an estimation of the species present in each trap, adults were collected, killed and stored at − 20 °C, after which each species was identified 58,59 . Adult D. melanogaster, D. simulans, D. hydei and D. subobscura were present in the traps.
Pupal size measurement and neutral lipid quantification. To measure pupal size, a similar procedure as described in Ref. 18 was followed. In short, each pupa was photographed individually using a camera linked to a stereomicroscope (Leica, SAPO). The total area of the pupal case was then measured using FiJi software (ImageJ v2.1.051; 60) . Quantification of the neutral lipid fraction (i.e., fat or triglycerides) was done using the protocol described in Ref. 61 . Briefly, pupae were dried in an oven at 60 °C for 3 days, after which dry weight was determined using a microbalance (Mettler Toledo, MT5). Each pupa was then placed into a glass tube with 4 ml of diethyl ether for 24 h. Pupae were then dried again at 60 °C for 3 days and pupal dry weight determined again, giving the neutral lipid-free dry weight. The absolute amount of fat (total neutral lipids in µg/pupae) was then obtained by subtracting the lipid-free dry weight from the lipid-containing dry weight. Ether extraction is an efficient method for fat quantification, as it extracts predominantly neutral lipids 62 , including triglycerides that represent lipids for energy storage 31 .
Statistical analyses. All analyses were performed with R version 4.0.2 63 . For laboratory-reared pupae, time to pupation, pupal size, and fat content were analyzed using generalized linear mixed-effects models (GLMM) with poisson (pupation time) or gamma (size and lipid content) error distributions. For all GLMMs, the fixed effect diet was analyzed using analysis of deviance with the "Anova" function from the "car" package 64 . Differences between diet groups were then identified using estimated marginal means comparisons (EMMs) using the "emmeans" function 65 . Vial number and pupa collection time were included as random factors (for pupation time, pupal size, and lipid content), as was the extraction run (for fat content). For each GLMM, models were simplified by removing random factors step by step, and compared using an anova with the "model.sel" function from the "MuMIn" package 66 . When models did not differ significantly and the AIC was smaller (delta AIC ≥ 2), the simplified model was kept based on methods of model simplification presented in Ref. 67 . The final model for the time to pupation included vial number as a random factor and the final model for fat content included pupae collection time and extraction run as random factors. No random factor was included in the final model for size. Final models are presented in Table 1.
The correlation between pupal size and fat content was analyzed using linear mixed-effects models (LMM) for laboratory and field-collected pupae separately. For laboratory pupae, the fixed factor diet was used as a covariable. Vial number, pupae collection time, and extraction run were included as random factors. The LMM was then simplified as described above for GLMMs, and only vial number was kept as a random factor in the final model ( Table 1). The LMM was further simplified for interactions between fixed factors. For field pupae, species was used as a co-variable (D. suzukii or "other species"). Flask number, pupae collection time and extraction run were included as random factors. The LMM was simplified as described above, but all random effects and interactions were included in the final model (Table 1). For both LMMs, statistical significance of each variable was determined using analysis of deviance. R 2 were then calculated for both LMMs using the "r.squaredGLMM" function from the "MuMIn" package. For mixed-effects models, R 2 comes in two types: marginal and conditional. Marginal R 2 represents the variance explained by the fixed effects of the model, while conditional R 2 is interpreted as a variance explained by the entire model, including both fixed and random effects. Both the marginal and conditional R 2 are reported.
Dietary conditions (nutrient content and/or availability) during development also affected time to pupation, which ranged on average from 6.64 (± 0.49) days for larvae that developed on the sugar-rich (2/1) medium to 17.73 (± 2.68) days for starved larvae following 2 days of feeding ( Table 2, GLMM, χ 2 = 77.196, df = 5, p value < 0.001). Starvation after 2 days of feeding and the no-sugar medium increased development time compared to controls (standard medium, 1/1). Starvation after 3 days of feeding, crowding, and the sugar-rich medium (2/1) did not induce significant changes in development time compared to controls (1/1), but the sugar-rich medium (2/1) shortened development time compared to crowding and starvation after 3 days (Table 2).
Pupal size and fat content are correlated in laboratory-reared and field-caught Drosophila pupae. We used the variation in pupal size and fat content produced by the different environmental conditions of developing D. melanogaster larvae to test for a positive correlation between the two traits. Pupal size and Table 2. Number of pupae formed (sample size for size measurement, fat content quantification, and time to pupation), and mean time to pupation (± sd) when development occurred on different diets. Different letters indicate significant differences based on estimated marginal means comparisons (p value < 0.05).

Discussion
In this study, we manipulated nutrient content and nutrient availability of developing D. melanogaster larvae to produce phenotypes that differ in pupal size and fat content. Starvation, dietary restriction, and crowding generally increased development time, but reduced pupal size and fat content. These conditions are indeed known to decrease body size and fat reserves in adult D. melanogaster 20,[68][69][70] , but also in other insects [71][72][73][74] . In this paper, the conditions that produced the smallest and leanest individuals were crowding and starvation. Crowding both decreases food availability (nutrients are consumed by conspecifics) and food quality (e.g., overconsumption of accumulated toxic waste produced by conspecifics in the food) 75 . For starvation, previous work showed that starvation of D. melanogaster larvae before 3 days of age (i.e., 70 h) can provoke major hormonal dysregulation that impedes reaching the minimal viable weight to initiate metamorphosis, thus leading to death 20,22 . In our study, indeed, only few individuals pupated under the harshest starvation condition (only 48 h of feeding after egg laying). Crowding and starvation are complex multifactorial stressors that can explain why these treatments led to the smallest pupae. The crowding conditions used in this study (300 eggs for 1 ml of food) represent a high density for developing individuals, but D. melanogaster larvae can develop in conditions of up to 1000 eggs for 1 ml of food. Survival, however, decreases dramatically under those conditions (with only 1.25% of individuals pupating 56 ). It would be interesting to test how even higher densities than the one used in this study would influence size and fat content of D. melanogaster.
Our data further showed no significant differences in size, fat content and development time between controls (1/1) and individuals developing on the sugar-rich medium (2/1). This is in contrast to several studies on D. melanogaster that showed that sugar-rich diets increased the fat content of adults 76,77 and larvae 55 , decreased body size 70,76 , and significantly increased time to pupation 52,77 . High sugar diets can further have deleterious effects on fitness, reducing fecundity 76 or egg to pupa viability 52 . Absence of significant differences between our sugar-rich medium (2/1) and controls (1/1) can result from the concentration of sugar that was used for our sugar-rich medium (10%). Indeed, in the study of Klepsatel et al. 76 , major differences in body size and fecundity were observed for Drosophila that developed on a medium containing 25% sugar. Furthermore, when compared with individuals from the no-sugar medium (0/1), 2/1 pupae were bigger, contained more fat and developed faster, showing that variation in sugar quantity had a significant impact on these life history traits. An increase in fat reserves in relation to sugar quantity in the diet can be expected, because excess sugars from the diet are metabolized through glycolysis and the tricarboxylic acid cycle to produce acetyl coenzyme A (acetyl-CoA). Acetyl-CoA is then converted to triglycerides that are stored in lipid droplets within adipocytes in the   31,78 . Overall, our findings, together with previous studies 20, [68][69][70] , suggest that the nutritional conditions experienced by D. melanogaster individuals during the larval stage directly affect the size and fat content of pupae and adults that can have major consequences for fitness. Pupal and adult size are intimately linked with fitness in insects 2,79 . In Aedes mosquitoes, for example, an increase in size of 25% leads to an increase in egg production of 100% 25,80 , while in D. melanogaster an increase in size of 43% can increase ovariole numbers by 290% 81 . In moths, larger individuals live longer, with a body size (wing length) increase of 25% leading to an increase of 79% percent in lifespan 82 . Similarly, in D. pseudoobscura a 24% increase in size can lead to a 32% increase in longevity 9 , while conversely decreasing sugar proportions in the diet decreases fat content of the flies by 26%, in turn decreasing median longevity by 122% 36 . Having large fat reserves can, however, also comes at a cost. D. melanogaster was found to show obesity-like phenotypes, with pathologies similar to those associated with obesity in humans, such as heart (dorsal vessel) failure, decreased endurance or metabolic dysregulation 50,55 . Despite these considerations, higher fitness in Drosophila is often associated with size and fat content (females with larger body size and higher fat reserves lay more eggs and live longer). This is clearly demonstrated by numerous studies where individuals artificially selected for a larger body size had higher fitness traits, including female fecundity 10,83 , longevity 84 or male reproductive success 10,84,85 . The positive relationship between body size and fitness is not limited to insects, but has also been found in vertebrates, including reptiles [86][87][88] and mammals 89,90 .
The main aim of this study was to use variation in size and fat content of Drosophila pupae (as a consequence of modifying nutritional conditions during developmental) to test the hypothesis that fat content is positively correlated with pupal size. We further collected wild pupae to observe if this correlation was also found in natural Drosophila populations. We observed that for field-caught Drosophila, fat content of 96% of D. suzukii pupae and 100% of pupae from other species were within the range of fat contents produced by our treatments (mainly induced by the nutrient-rich media, i.e., 2/1 and 1/1; see Fig. 3). Validating our hypothesis, pupal fat content was indeed positively correlated to pupal size both in laboratory-reared D. melanogaster and field-caught Drosophila species. Other works however, showed that some conditions can promote an opposite trend. Indeed, Kristensen et al. 44 showed that flies developing on a protein enriched medium were larger than controls (measured as dry www.nature.com/scientificreports/ weight), but had fewer lipid reserves (in % of fly body mass). In our study, we increased the proportion of sugar in the diet, but not the proportion of protein. It would be interesting to test how a gradient of protein concentrations can affect the relationship between body size and total fat content in D. melanogaster.
In this study, we did not separate male and female pupae. Yet, we observed a strong relationship between fat content and size, with low variability for both laboratory and field pupae, as highlighted by the high conditional R 2 (0.85 and 0.78, respectively). In arthropods, females tend to have higher fat reserves than males, but females also generally have a larger body size; hence the relationship between size and fat reserves is expected to be similar for males and females 30 . Pupal size, an easy and reliable measurement, is therefore a robust predictor of fat content for both laboratory-reared and field-caught Drosophila species. In insects, several body measurements, such as head width 91 , tibia length 5 , wing length 92 , thorax or abdomen length 93 , and pupal size 17 are used as a proxy for body size, fat content, and fitness. As insect pupae are easy to handle in general, measuring pupal size represents a convenient and non-invasive method to estimate fat content and potentially fitness. By linking pupal size to fat content, our results can be of interest for future eco-evolutionary and physiological studies, because this method allows to estimate an individual's energetic reserves with the option to use the same individual for further experimentation and life history measurements.
Drosophila larvae and pupae are hosts to numerous parasitic wasp (i.e., parasitoid) species 94 . During development, a parasitoid consumes fat from only a single host insect and fat stores are often not replenished during adult life 95,96 . Host resources available for developing parasitoids can thus in turn have major consequences for adult wasp life history traits 97 . Measuring pupal size of Drosophila, therefore, offers the possibility to estimate the amount of fat available for developing parasitoids 98 . Our data can also be of interest from an applied perspective, as the spotted wing drosophila (D. suzukii) is a major pest of red berries. Several studies are ongoing to develop biological control methods to fight this pest, such as the sterile or incompatible insect techniques [99][100][101] . The success of these techniques, which require mass release of sterile/incompatible males in infested areas, relies on the quality of released males 102 . In mass rearing facilities, measurements are made at several critical points to check insect quality. Our data shows that pupal size is a reliable estimate of fat content of Drosophila flies, including D. suzukii, and could, therefore, represent a good quality measurement of mass-produced individuals.

Data availability
All data used on this article are available on demand from the corresponding author. www.nature.com/scientificreports/ Reprints and permissions information is available at www.nature.com/reprints.
Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.