Intergenerational paternal effect of adult density in Drosophila melanogaster

Abstract Notwithstanding recent evidences, paternal environment is thought to be a potential but unlikely source of fitness variation that can affect trait evolution. Here we studied intergenerational effects of males’ exposure to varying adult density in Drosophila melanogasterlaboratory populations. We held sires at normal (N), medium (M) and high (H) adult densities for 2 days before allowing them to mate with virgin females. This treatment did not introduce selection through differential mortality. Further, we randomly paired males and females and allowed a single round of mating between the sires and the dams. We then collected eggs from the dams and measured the egg size. Finally, we investigated the effect of the paternal treatment on juvenile and adult (male) fitness components. We found a significant treatment effect on juvenile competitive ability where the progeny sired by the H‐males had higher competitive ability. Since we did not find the treatment to affect egg size, this effect is unlikely to be mediated through variation in female provisioning. Male fitness components were also found to have a significant treatment effect: M‐sons had lower dry weight at eclosion, higher mating latency, and lower competitive mating success. While being the first study to show both adaptive and non‐adaptive effect of the paternal density in Drosophila, our results highlight the importance of considering paternal environment as important source of fitness variation.


| INTRODUC TI ON
Parental environment has the potential to influence offspring traits and fitness through intergenerational effects (and more stable transgenerational effects, see Dias & Ressler, 2014 for the distinction between trans and intergenerational effects).While it can potentially pass on deleterious effects of different components of the environment to the following generation (Yehuda et al., 2000), intergenerational effect can also be adaptive, especially under fluctuating environment (Bonduriansky & Day, 2009). Among the myriad components of an organism's ecology, few factors are as variable as density and nutritional availability. Both have been recently found to have intergenerational effects, especially through the maternal route (i.e., maternal effect), in a wide variety of organisms (Mousseau & Fox, 1998a, 1998b. There is a growing body of evidence showing the importance of the intergenerational effect of paternal nutrition, social experience and density on fitness related traits of the offspring (Adler & Bonduriansky, 2013;Crean, Dwyer, & Marshall, 2013;Friberg, Stewart, & Rice, 2012). However, the prevalence and adaptive significance of such paternal effect is yet to be ascertained.
There are many reports of environment dependent maternal effect mediated through variation in maternal provisioning in egg/ offspring (Mousseau & Fox, 1998a, 1998bRossiter, 1996). For example, females living under high density may suffer from adverse effects of crowding (such as, malnutrition) and may therefore struggle to allocate resources in maternal provisioning-either in the form of stored resources in egg or lactation, which in turn may lead to poor quality progeny (Christian & Lemunyan, 1958).
Alternatively, females raised in high density may strategically produce fewer eggs/progeny while investing more resources (e.g., yolk) in each of them-thereby giving the progeny a better start for the impending challenges of crowding (Holbrook & Schal, 2004;Mitchell & Read, 2005;Prasad, Shakarad, Rajamani, & Joshi, 2003;Vijendravarma, Narasimha, & Kawecki, 2010). Generally, under fluctuating environmental conditions, such parental ability to optimize offspring phenotype has been conjectured to be adaptive (Bonduriansky & Day, 2009;Kuijper & Hoyle, 2015). For example, Guppy (Poecilia reticulata) females were found to produce larger offspring (a) under food limitation (Reznick & Reznick, 1993) and (b) when they experienced high level of competition-priming the offspring for better competitive ability (Bashey, 2006). The larger eggs produced by Drosophila melanogaster females that grew in nutritionally impoverished food, survive (egg-toadult survivorship) better in impoverished food and give rise to smaller adults (Vijendravarma et al., 2010). In contrast, Valtonen, Kangassal, Pölkki, and Rantala (2012) found that D. melanogaster females grown on impoverished food produced larger offspring (adult) compared to those grown on nutritionally rich food. Note that many of the maternal effects discussed above are mediated through variation in resource provisioning by mothers.
Not surprisingly, most of the reports of environment dependent paternal effect (intergenerational and transgenerational) are from animals with paternal provisioning through nuptial gift transfer to the females (Dussourd et al., 1988;Gwynne, 1988;Smedly & Eisner, 1996;Vahed, 1998;Zeh & Smith, 1995). However, it is only recently that studies have started addressing the presence of similar paternal effects in species without paternal provisioning. For example, in one of the first such explicit studies, female Neriid flies (Teleostylinus angusticollis) raised on richer diet were found to produce larger eggs and offspring that developed faster, while males raised on richer diet sired larger offspring with better survival rate, especially under resource scarcity (Adler & Bonduriansky, 2013;Bonduriansky & Head, 2007).Further, in a solitary Ascidian, Styelaplecata, males were found to produce offspring with phenotype corresponding to the population density experienced by the father (Crean et al., 2013). In fruit flies (D. melanogaster) Valtonen et al. (2012) reported that fathers fed on poor quality diet sire larger sons. Paternal experience of the intensity of competition (assessed by the number of co-inhabitant rival males) was found to affect reproductive behavior (duration of copulation) of male offspring in D. melanogaster . In Desert locusts (Schistocerca gregaria), Islam, Roessingh, Simpson, and McCaffery (1994) showed paternal crowding to have a significant impact on hatchling coloration and nymph behavioral traits. Paternal experience of ambient temperature was also found to affect offspring fecundity in D. melanogaster (Huey, Wakefield, Crill, & Gilchrist, 1995). Low temperature was found to affect offspring phenotype in two other species of Drosophila-D. simulans (Watson & Hoffmann, 1995) and D. serrata (Magiafoglou & Hoffmann, 2003). Thus, there is ample evidence showing environment dependent paternal effect. In addition to affecting viability, such paternal effect has been shown to affect progeny reproductive performance and hence is likely to be key player in sexual selection (for example, see Bonduriansky & Head, 2007). However, such data are far from being plenty.
Here, we investigated the effect of paternal experience of population density on progeny fitness components, including male mating behavior in D. melanogaster laboratory adapted populations. As discussed previously, paternal effect has already been reported in these  and other populations of D. melanogaster, establishing them as a relevant system to investigate the paternal effect and its consequences on Darwinian fitness (William et al., 2006). Further, laboratory adapted populations of D. melanogaster have been used to investigate the fitness consequence of a plethora of environmental parameters, including population density. Fruit flies naturally grow in ephemeral resource patches, such as rotting fruits and vegetables. Crowding in transiently available rich patches is expected to be a key component of their natural ecology. Density of adults in a resource patch not only determines the extent to which individuals must compete for food and limited space (e.g., oviposition substrate) but also for suitable mates. Increase in density also leads to an increase in the probability of disease transmission (Barnes & Siva, 2000). In essence, density often determines the nature and intensity of selection acting on a population and has been studied within the broader premises of density dependent selection (MacArthur   Wilson, 1967;Mueller, 1997;. Much of the existing literature investigated adaptation to increased (but stable) juvenile or adult density, using experimental evolution on laboratory populations of D. melanogaster (Mueller, Guo, & Ayala, 1991;Mueller & Sweet, 1986;Nagarajan, Natarajan, Jayaram, & Joshi, 2016;Sarangi, Nagarajan, Dey, Bose, & Joshi, 2016;. However, little is known about adaptation to fluctuating density. Intergenerational and transgenerational effects, if used by the parents to optimize offspring phenotype, can be of adaptive value if density fluctuation across generation is, at least to some extent, predictable. Interestingly, these experimental evolution studies reported "rapid" adaptation to "crowding". Though evidences unequivocally showed the genetic changes associated with such adaptation, non-genetic parental effects (trans and intergenerational) may, in addition, account for the "rapid" adaptation (Bonduriansky & Day, 2009). However, this idea has not been tested-an existing lacuna in the literature, which we intend to fill to some extent.
To investigate the paternally transmitted intergenerational effect of varying density, we subjected males to three adult density treatments and then allowed them to sire progeny by mating the treated males to untreated dams. We then assessed the effect of the paternal adult density (hereafter, referred to as paternal density) treatment on progeny fitness components in juvenile (juvenile competitive fitness) and adult stages (males: mating ability, mating latency, copulation duration, courtship frequency, competitive mating success). We found the paternal density treatment to have significant intergenerational effect on both juvenile and adult fitness components.

| MATERIAL S AND ME THODS
All the experiments were done using a set of laboratory adapted populations of D. melanogaster-BL. Full laboratory history of these populations can be found in (Nandy, Dasgupta, Halder, & Verma, 2016). Briefly, these are a set of five replicate populations (BL 1-5 ) maintained on standard Banana-Jaggery-Yeast food, under 14-day discrete generation cycle at 25°C ambient temperature, 60%-80% relative humidity, with population size ~2,800. Larval density is

| Paternal treatment
Sires and dams were generated from a BL population. The broad design of the protocol followed to generate the experimental flies F I G U R E 1 The design of the assay. The schematic diagram shows the design of the entire study, which spanned two generations. Treatment (Normal [N], Medium [M] and High [H] adult densities) was given in the paternal generation. Untreated dams were mated to the treated sires, followed by the collection of eggs from the dams. Assays were done with the eggs and the offspring emerging out of the eggs. Some eggs were subjected to mixed culture (along with competitor eggs) and juvenile competitive fitness vials were set up. Some eggs were cultured as monocultures (without any competitor eggs)-male progeny emerging from these vials were used for further assays, such as, mating ability, mating latency, copulation duration, competitive mating success and courtship frequency  Figure 1. To generate the experimental sires and the dams, eggs were collected from a BL population and cultured under standard density (i.e., 70 per 6-8 ml food per vial). 100 such vials were set up, of which 65 were used to collect the sires (=sire-vials) and the remaining 35 for dams (dam vials). Dams were collected as virgins and held in single sex vials at a density of 25 per vial with ad-lib food until the day of the sire-dam mating (see below). In the sire-vials, all the flies were allowed to eclose. These flies were used to set up three adult density treatments-normal (N: 70 individuals per vial), medium (M: 140 individuals per vial) and high (H: 210 individuals per vial; see Figure 2). 10 vials were set up for each of the treatments, using flies that were approximately 1-day old. These vials were left undisturbed for 2 days, following which males from them were separated and used as sires in the subsequent step. Here and elsewhere throughout the study, all the fly sorting, including collection of virgins, were done under light CO 2 -anesthesia, unless mentioned otherwise.

| Sire-dam mating
Following the 2-day-long conditioning, 25 males were randomly isolated from each adult density treatment vials, to be used as sires.
They were then combined with dams (see previous section) in fresh food vials (25 sires + 25 dams in a vial) and allowed to interact for 90 min, which is sufficient time for a single round of mating. This method of ensuring single round of mating has been previously used (Nandy, Joshi, Ali, Sen, & Prasad, 2012). In addition, mating was visually observed. Occasionally, in some vials, a small number of females failed to mate within this time. We did not make any attempt to remove them. These un-mated females either mated with an already mated male after a while (late mating) or remained un-mated. Most males secured a single mating, while some very small number (those which mated with the un-mated females mentioned earlier) may have secured more. The number of such late matings (and hence, male re-mating) was very small, and therefore very unlikely to have any perceivable impact on the subsequent assays. Further, the females in this system usually do not re-mate within such short span (i.e., 90 min) unless the first one was a failed mating, which is very rare in our populations. Therefore, by following this protocol, we generated singly inseminated females (average number per vial ~25).
10 mating vials were set up per density treatment. After mating, the sires were discarded and the already inseminated dams from all 10 vials of a treatment (i.e., a total of 250 females) were transferred to a 2 L plastic cage with food smeared with ad-lib quantity of live yeast.
Three such cages were thus set up-one for each density treatment.
After 2 days, eggs were collected from these cages to set up the remainder of the experiments. To collect the eggs, a fresh food plate was introduced in the cage. The dams were allowed a short window (2-3 hr) for oviposition. Using a fine brush, eggs were counted on to a fine agar strip, which was then transferred to the culture vials (see below). These eggs are hereafter referred to as treatment eggs.

| Measurement of egg size
To test if the sires influenced the size of the eggs laid by the dams (Pischedda, Stewart, Little, & Rice, 2010), a subset of these eggs were frozen at −20°C and their size was measured. For this purpose, eggs were mounted on a glass slide on their dorsal side and photographed using Nikon Stereo-zoom trinocular microscope (SMZ745T) and the area of the two-dimensional elliptical outline of the eggs were measured in ImageJ, software. This area was taken as a proxy for the size of each egg. A given egg was measured thrice and the average of these three measurements was taken as the unit of analysis. 50 eggs per treatment were measured for this purpose.

| Experiment 1: Juvenile fitness assay
Egg-to-adult survivorship was taken as a measure of Juvenile fitness. Survivorship of the treatment eggs were measured against a back ground of a common competitor (BL st ) under two conditionscrowded (C: 150 larvae per 1.5 ml food in each vial) and un-crowded (UC: 70 larvae per 6 ml food in each vial). During the assay, treatment eggs generated in the previous step were cultured with eggs from common competitors in the ratio 1:4 (C: 30 targets, 120 competitors; UC: 14 targets, 56 competitors). These common competitors were collected from an untreated BL st stock. On completion of development, it was possible to identify the target progeny from the competitor progeny based on eye color-progeny of the competitors were scarlet-eyed, whereas the target progeny was red-eyed. 10 juvenile competition vials were set up for each of the three treatments (viz., N, M and H) and two assay conditions (i.e., 10 as C and 10 as UC for each treatment). These vials were left undisturbed until adult emergence was complete (12th day post-egg deposition). The adults were sorted based on eye color and counted. Juvenile fitness score (w) was calculated for each vial following the formula: w = number of red eyed progeny observed number of red eyed progeny expected F I G U R E 2 A typical mating arena having five mating pairs and the three treatment groups-High (H), Medium (M) and Normal (N). The flies are held in glass vials, on banana-jaggery-yeast food. As shown here, food given to them was ad-lib, with ample space for the flies to move around The number of red-eyed progeny expected was 14 and 30 for UC and C-assay conditions respectively.

| Experiment 2: Assay for behavior and fitness of the sons
To investigate the effect of the treatment on the male progeny, the treatment eggs were cultured in food vials in the usual density (i.e., For all values of x, until, n x−2 ≤ n x .
For all values of x, until, n x−2 ≥ n x .
Occasionally, some females did not mate within 1-hr long observation. These flies were excluded from the analysis. Similarly, some males also failed to secure mating. In vials having such an unsuccessful male, a mating was recorded much later-when one of the successful males finished its first mating and then initiated a second one with the un-copulated female. Such late copulations were also excluded from the analysis. Mating ability (MA) is measured as the proportion of the sons successfully copulated. MA was calculated for every single vial.
(c) Courtship frequency was quantified for the 3-day-old (post-eclosion) sons of the three paternal density treatments by setting up similar mating vials as described in the previous section.  (Bastock & Manning, 1955;Sokolowski, 2010). Any of the above-mentioned courtship behaviors, displayed by the five males in each vial was counted as one. The total number of independent male to female courtship displays was counted within the observation window (Bedhomme, Prasad, Jiang, & Chippindale, 2008;Nandy et al., 2013). The treatment identities were unknown to the observers to avoid observer bias. (d) In the fourth assay, another set of males were similarly collected and held, to be used for quantifying their mating success under competitive condition (CMS, Competitive mating success).
This was done by setting up mating vials with five 3-day-old target males, five competitor males (BL st ) and five virgin females (BL st ).
Ten such mating vials were set up for each of the three treatments.
After allowing a single round of mating (i.e., for 90 min) for all the females in a mating vial, the females were individually transferred to oviposition test tubes (12 mm diameter × 75 mm height) with ample food. The females were allowed to oviposit for 18 hr.
Following oviposition, the females were discarded and the tubes were retained to allow the progeny to develop and eclose. For each female, the identity of their mate (whether target/competitor) was ascertained by observing the eye color of the progeny.
Progeny sired by target males were red-eyed whereas those sired by competitors were scarlet-eyed. For a given vial, average CMS of the five target males in the vial was calculated as the proportion of the females mated to target males (i.e., produced red-eyed offspring).

| Experimental replications and data analyses
The entire study was carried out in three randomized blocks, using three different BL populations-BL 1 , BL 3 , and BL 5 . The blocks were handled on separate days. Number of replications within each block has been mentioned in the previous sections along with the assay design. Except for the egg size and dry body weight assay, all the experimental replication was done at the level of assay mating vials or juvenile competition vials. All the assays had 10 replicate vials.
Vial means were used as the unit of analysis. For egg size assay, size of each egg was used as the unit of analysis. For dry body weight, weight of groups of five individuals was used as the unit of analysis. Data were analyzed using mixed-model analysis of variance (ANOVA). Block was treated as random factor, while paternal density treatment and assay density (wherever applicable) were treated as fixed factors. Multiple comparisons were done using Tukey's HSD.
All the analyses were done in Statistica, version 10 (Statsoft, Tulsa, OK, USA).

| RE SULTS
Variation in size of the eggs represents variation in maternal provisioning. The effect of the paternal density treatment on size of the eggs produced by the dams was not significant ( Note. Paternal density and assay density (wherever applicable) were considered as fixed factor, while block as random factor. All tests were done considering α = 0.05 and significant p-values are mentioned in bold face.
of the H-treatment was only evident under larval crowding, that is, C-assay density (Figure 3), indicating competitive superiority of the H-juveniles. However, the paternal treatment × assay density interaction was marginally non-significant (Table 1).
We only quantified the effect of the paternal density on male offspring. We found a significant effect of the treatment on dry body weight, ML and CMS (Table 1, Figure 4). Multiple comparisons using Tukey's HSD indicated that dry body weight of the M-sons were significantly less than that of the N-sons, with M-sons having 8.7% lower mean dry body weight. The difference between the dry body weight of the H and N-sons was not statistically significant. Hence the M-sons were significantly smaller compared to the other two treatments. In the mating assay, though we found some males to fail in acquiring mating, there was no effect of the treatment on mating that of the N-sons. This is however, not due to a reduced courtship performance by the M-sons as we found the effect of the treatment on CF (mean ± SE, N: 6.7 ± 0.6; M: 6.8 ± 0.6; H: 7.6 ± 0.8) to be nonsignificant. We also did not find any effect of the treatment on CD (mean ± SE, minutes, N: 18.6 ± 0.4; M: 17.6 ± 0.4; H: 17.9 ± 0.4), potentially indicating the lack of the treatment effect on post-copulatory traits of the sons (Table 1).

| D ISCUSS I ON
Given that very few studies have shown the effect of paternal environment on offspring fitness components, there were two main objectives of the present study-(a) to assess if paternal exposure to varying population density affected progeny traits; if yes, then (b) to evaluate the adaptive significance of such effect. The results clearly showed that at sufficiently high density, males had an adaptive paternal effect on juvenile competitive fitness. As we did not find any effect of our treatment on size of the eggs produced by the dams, such paternal effect is unlikely to be mediated by variation in provisioning by the females. We further show that at intermediate density, males sire smaller sons which are inferior in acquiring mates. Interestingly, such maladaptive effect of paternal density on offspring adult fitness was not detected at high density.
In holometabolous insects like fruit flies, juvenile (larva and pupa) survival constitutes one of the most important components of fitness . In addition, juvenile ecology may also have a major effect on the life-history and fitness components of the adult stage (Heat shock: Khazaeli, Tatar, Pletcher, & Curtsinger, 1997;cold shock: Singh, Kochar, & Prasad, 2015;Singh & Prasad, 2016;crowding: Joshi & Mueller, 1988;Sarangi et al., 2016;. The observed paternal effect on juvenile competitive fitness therefore is extremely consequential. Some relatively recent studies have pointed out that evolving parental ability to optimize offspring fitness related traits can be an adaptation to ecological challenges (Galloway & Etterson, 2007), including crowding (Crean et al., 2013). Given that fruit fly natural ecology regularly involves adult and larval crowding, the observed paternal effect on juvenile competitive fitness can indicate males' adaptation to crowding. Interestingly, we observed the paternal effect on juvenile competitive fitness, only at the highest density, which may indicate a certain threshold density beyond which such paternal effect starts affecting offspring traits. In addition, when assayed under un-crowded condition the progeny from the three sire treatments do not show any measurable difference in their egg-to-adult survival. This suggests that the juvenile competitive ability rather than baseline juvenile viability was affected by the treatment. Since a number of traits (e.g., feeding rate, waste tolerance, development time etc.) affect juvenile competitive ability in these flies, it will be interesting to F I G U R E 3 Effect of the paternal density treatment on juvenile competitive fitness, under crowded and un-crowded assay conditions. Target eggs (eggs produced by dams mated to treatment sires) were cultured with competitor eggs (eggs produced by untreated females and males) in juvenile competition vials-under un-crowded and crowded conditions. Proportion of the target progeny successfully emerging as adults is considered as the measure of juvenile competitive fitness. Black, blue and red color coding represent the progeny of Normal (N), Medium (M) and High (H) density treatment males respectively. The H-progeny were found to have higher juvenile competitive fitness compared to N and M-progeny (represented by *asterisk), only when assayed under crowded condition. The entire experiment was done following a randomized block design and the data were analyzed using threefactor mixed-model ANOVA with paternal treatment and assay condition as fixed factors and block as random factor find out the trait responsible for better competitive ability of the H-sons in our study.
In a wide range of species including D. melanogaster, maternal exposure to high density or poor nutrition has been found to affect offspring fitness components Valtonen et al., 2012;Vijendravarma et al., 2010). Such effects can either be beneficial (Allen, Buckley, & Marshall, 2008;Bashey, 2006;Gorbi, Moroni, Sei, & Rossi, 2011;Mitchell & Read, 2005;Vijendravarma et al., 2010) or detrimental (Meylan, Clobert, & Sinervo, 2007) Sarangi et al., 2016 and the references therein for an updated review). Therefore, one possibility is that genetically superior males, which are better at surviving under high density, may produce offspring which are better both as juveniles, explaining at least part of our observations. Though larval competitive ability is known to respond to experimental evolution, indicating heritable genetic variation (Mueller, 1997;, such heritable variation is very unlikely to have led to the observed treatment effect on juvenile competitive fitness. This is because (a) in our assay, we recorded very little mortality in males during the treatment, indicating negligible hard selection. In addition, we also ensured that there was no soft selection by randomly picking the set of males from the treatment vials to use them as sires. Further, we allowed the sires and the dams to mate only once by allowing them a limited window of time to interact after being put together in mating vials. (b) Even if there was selection in the current experimental design, the selection is likely to be weak (see Materials and Methods section). Such weak selection is unlikely to explain the observed differences in some of the traits (viz., 8.9% increase in juvenile competitive fitness, 35% increase in mating latency), especially within one generation.
Alternatively, males may alter maternal provisioning and thereby indirectly affect offspring fitness components Vijendravarma et al., 2010). We, however, did not find any measurable difference in the size of the eggs produced by females mated to the males belonging to the three treatments, making variation in maternal provisioning an unlikely explanation. Therefore, although F I G U R E 4 Effect of the paternal density treatment on male offspring. (a) Dry body weight at eclosion: five flies were weighted together to nearest 0.01 mg. This was then used as the unit of analysis; (b) Mean mating latency (time taken by a virgin malefemale pair to start copulation): mean ML was calculated for five males in a vial following the algorithm given in the Materials and Methods section. This was done for all the mating vials in the assay. These values were then used as the unit of analysis; (c) Competitive mating success (CMS): CMS values were calculated for each vial having five target males as the proportion of females mated to target males in these assay vials. These values were then used as the unit of analysis. The blue broken line indicates the expected value of CMS if there is no mating bias. Black, blue and red color coding represent data from the progeny of N, M and H-males respectively. Treatments not sharing common alphabet were found to be significantly different from each other. The entire experiment was done following a randomized block design and the data were analyzed using three-factor mixed-model ANOVA with paternal treatment and assay condition as fixed factors and block as random factor sire effect on the quality of the eggs produced by the females cannot be completely ruled out without a more detailed qualitative analysis of the eggs, our results tentatively point at non-genetic paternal effect (Bonduriansky & Day, 2009) as the potential cause behind the observed effect of the treatment. Interestingly, a recent study on D. melanogaster has shown paternal effects to have important consequences on the expression of an array of genes in sons (Zajitschek, Zajitschek, & Manier, 2017, also see the corresponding correction).
In addition, Garcia-Gonzalez and DowlingD (2015)  While we found adaptive paternal effect on juvenile performance, adult performance however, was found to have a significant maladaptive effect of paternal density. Males that experienced intermediate density were found to sire sons which (a) are smaller, (b) take longer time to start mating and (c) have lower mating success.
Since we did not find any effect of the treatment on courtship frequency, reduced mating success and increased mating latency was a likely outcome of females' reluctance to accept relatively smaller males as their mates, a known fitness consequence of reduced size in Drosophila males (Jagadeeshan, Shah, Chakrabarti, & Singh, 2015;Partridge, Ewing, & Chandler, 1987). Body size has been reported to be affected by intergenerational paternal effect in another Dipteran-Telostylinus angusticollis (Bonduriansky & Head, 2007).
Unlike the maladaptive effect found in our study, the paternal effect on body size reported by Bonduriansky and Head (2007) was adaptive, especially under certain prevailing conditions. In D. melanogaster, the effect of body size on different fitness components has been found to be context specific (Lefranc & Bundgaard, 2000;Morimoto, Pizzari, & Wigby, 2016;Pitnick, 1991). Thus, although in the context of our experiment, the observed body size reduction in our study appears to be maladaptive, a detailed investigation where offspring fitness is measured under varying adult density is necessary to better understand the fitness consequences. An adaptive paternal effect theory would predict higher fitness of the offspring, particularly when offspring conditions match the paternal conditions. Interestingly in our study, the paternal (maladaptive) effect was found only at intermediate density and not in the high density treatment. However, at this point it is difficult to suggest any reason for such specific expression of the paternal effect at intermediate density.
As variation in population density and crowding related ecological challenges are common in almost all organisms, including fruit flies, paternal effect of the nature reported here is important to understand. Though paternal ability to optimize offspring traits is likely to be adaptive, especially under fluctuating environment, the results reported here show that paternal effect can be both adaptive and maladaptive. To the best of our knowledge this is the first evidence of the effect of paternal density on juvenile and adult fitness components in D. melanogaster. Importantly, our results emphasize the importance of considering paternal effect as a source of variation in fitness related traits. The full impact of such paternal effect in the evolution of life-history traits and the underlying mechanisms are emerging as an important topic of discussion, which is likely to see an increasing attention in years to come.

ACK N OWLED G M ENTS
We are immensely thankful to N. G. Prasad (IISER Mohali) for sharing the fly populations, using which the BL populations were created and also for his critical comments on a previous version of this manuscript. We are also thankful to Syed Zeeshan Ali and Vanika Gupta for their critical comments on a previous version of the manuscript.

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

AUTH O R CO NTR I B UTI O N S
BN, PD and SS conceived the ideas and designed the assays. PD, SS, AAD, TV, and to a lesser extent BN performed the assays and collected the data. Data analysis was primarily done by BN and to some extent, by PD. While BN and PD led the writing of the manuscript, SS and TV provided important assistance and inputs.