1. Ecology
  2. Evolutionary Biology

Thermal phenotypic plasticity of pre- and post-copulatory male harm buffers sexual conflict in wild Drosophila melanogaster

  1. Claudia Londoño-Nieto  Is a corresponding author
  2. Roberto García-Roa
  3. Clara Garcia-Co
  4. Paula González
  5. Pau Carazo
  1. Ethology Lab, Cavanilles Institute of Biodiversity and Evolutionary Biology, University of Valencia, Spain
  2. Department of Biology, Lund University, Sweden
https://doi.org/10.7554/eLife.84759
Share this article
Cite this article
  1. Claudia Londoño-Nieto
  2. Roberto García-Roa
  3. Clara Garcia-Co
  4. Paula González
  5. Pau Carazo
(2023)
Thermal phenotypic plasticity of pre- and post-copulatory male harm buffers sexual conflict in wild Drosophila melanogaster
eLife 12:e84759.
https://doi.org/10.7554/eLife.84759
  2. Download BibTeX
  3. Download .RIS

Abstract

Strong sexual selection frequently leads to sexual conflict and ensuing male harm, whereby males increase their reproductive success at the expense of harming females. Male harm is a widespread evolutionary phenomenon with a strong bearing on population viability. Thus, understanding how it unfolds in the wild is a current priority. Here, we sampled a wild Drosophila melanogaster population and studied male harm across the normal range of temperatures under which it reproduces optimally in nature by comparing female lifetime reproductive success and underlying male harm mechanisms under monogamy (i.e. low male competition/harm) vs. polyandry (i.e. high male competition/harm). While females had equal lifetime reproductive success across temperatures under monogamy, polyandry resulted in a maximum decrease of female fitness at 24°C (35%), reducing its impact at both 20°C (22%), and 28°C (10%). Furthermore, female fitness components and pre- (i.e. harassment) and post-copulatory (i.e. ejaculate toxicity) mechanisms of male harm were asymmetrically affected by temperature. At 20°C, male harassment of females was reduced, and polyandry accelerated female actuarial aging. In contrast, the effect of mating on female receptivity (a component of ejaculate toxicity) was affected at 28°C, where the mating costs for females decreased and polyandry mostly resulted in accelerated reproductive aging. We thus show that, across a natural thermal range, sexual conflict processes and their effects on female fitness components are plastic and complex. As a result, the net effect of male harm on overall population viability is likely to be lower than previously surmised. We discuss how such plasticity may affect selection, adaptation and, ultimately, evolutionary rescue under a warming climate.

Editor's evaluation

This study has important implications for the impact of sexual conflict on population viability under different temperatures. The authors provide compelling evidence that male harm to females in sexual conflict can be reduced as a function of temperature within the optimal reproductive range of a species. The results have implications for the likelihood of the evolutionary rescue of species facing the climate crisis.

https://doi.org/10.7554/eLife.84759.sa0

Introduction

Females and males share the common goal of siring offspring. This central tenet of sexual reproduction enforces a certain degree of cooperation between the sexes. However, anisogamy frequently leads to distinct sex roles and thus general asymmetries in the reproductive evolutionary interests of females and males, which can in turn result in diverging intensity and form of sexual selection across the sexes (Arnqvist and Rowe, 2005; Chapman et al., 2003a; Janicke et al., 2016; Winkler et al., 2021). This phenomenon, termed sexual conflict, favors traits in one sex that might be costly for the other (Parker, 1979), and can thus lead to antagonistic female-male coevolution (Arnqvist and Rowe, 2005). Sexually antagonistic co-evolution has received much attention and is recognized as a fundamental process in evolution due to its role in shaping male and female adaptations (Bonduriansky et al., 2008), in contributing to drive reproductive isolation and speciation (Arnqvist et al., 2000; Bonduriansky, 2011; Bonduriansky and Chenoweth, 2009; Gavrilets, 2014), and as a major determinant of population demography (Kokko and Brooks, 2003; Bonduriansky and Chenoweth, 2009; Berger et al., 2016). Specifically, sexual conflict has been shown to have profound consequences for female fitness and population growth when it favors male reproductive traits that increase male intra-sexual competitive ability at the expense of harming females (i.e. male harm, Crudgington and Siva-Jothy, 2000; Gómez-Llano et al., 2023; Wigby and Chapman, 2004).

Harmful male adaptations are widespread and incredibly diverse and sophisticated across the tree of life (Arnqvist and Rowe, 2005). For example, male harassment of females during pre-copulatory competition for mating has been documented in myriad vertebrate and invertebrate species (Gómez-Llano et al., 2023), driving antagonistic female-male co-evolution in a host of behavioral and morphological traits (Arnqvist and Rowe, 2005). Male harm adaptations in the context of post-copulatory competition are similarly widespread in invertebrates, featuring (amongst others) toxic ejaculates (Wigby and Chapman, 2005), love darts (Koene and Schulenburg, 2005), and a range of male adaptations for traumatic insemination that range from genital ablation to spiny penises (Crudgington and Siva-Jothy, 2000; Lange et al., 2013). Importantly, beyond driving female and male phenotypes and associated diversification processes, male harm generally leads to a ‘reproductive tragedy of the commons’ that can substantially impact population demography by depressing net female productivity (Arnqvist and Tuda, 2010; Berger et al., 2016; Holland and Rice, 1999; Rankin et al., 2011), and even facilitate population extinction (Le Galliard et al., 2005). Understanding what factors underlie male harm evolution, its diversity in form, strength, and outcomes, is thus a main concern in evolutionary biology.

Despite a growing number of studies in the field of sexual conflict, most have been conducted under uniform laboratory conditions, frequently in populations adapted to stable environments for hundreds of generations (Chapman et al., 2003b; Hopkins et al., 2020; Wigby and Chapman, 2004). In contrast, recent research has highlighted the role of ecology in shaping the evolution of traits under sexual conflict (Arbuthnott et al., 2014; García‐Roa et al., 2019; MacPherson et al., 2018; Perry and Rowe, 2018; Yun et al., 2017), including habitat complexity (Malek and Long, 2019; Miller and Svensson, 2014; Myhre et al., 2013), nutritional status (Fricke et al., 2010), or sex ratio and population density (Chapman et al., 2003a). For example, Gomez-Llano et al., 2018 recently showed that conspecific densities and the presence of heterospecifics modify the intensity and outcome of sexual conflict in the banded demoiselle (Calopteryx splendens), and the spatial complexity of the environment in which mate competition occurs mediates how sexual conflict operates in fruit flies (Yun et al., 2018). The incorporation of more realistic ecological scenarios in the study of sexual conflict is thus a key avenue to disentangle the evolution of male harm, and its consequences for population viability (Cornwallis and Uller, 2010; Fricke et al., 2009; Plesnar-Bielak and Łukasiewicz, 2021).

Temperature is recognized as a crucial abiotic ecological factor due to its impact on life history traits and physiological and behavioral responses (De Lisle et al., 2018; Kim et al., 2020; Miler et al., 2020; Monteiro et al., 2017). Furthermore, temperature varies in nature widely within and across spatiotemporal scales (e.g. daily, inter-seasonal, and intra-seasonal variation). Consequently, it may have short, medium, and long-term effects on organism phenotypes that can impact many different aspects of its reproductive behavior (e.g. sex-specific potential reproductive rates, operational sex ratios, density, etc.; García-Roa et al., 2020). In fact, a recent meta-analysis suggests that temperature may have a sizeable effect on sexual selection processes even when fluctuations occur well within the normal range of temperature variation for the studied species (García-Roa et al., 2020). This latter finding is particularly relevant given that we know almost nothing about how average temperature fluctuations, such as those experienced by wild populations during their reproductive season, affect male harm and sexual selection at large.

Our aim was to contribute to fill this gap in knowledge by studying how male harm responds to temperature variation that mimics average fluctuations that are normal during the reproductive season, using Drosophila melanogaster flies sampled from a wild population. D. melanogaster is an ideal subject for this study because it exhibits high levels of male-male competition, has well-characterized pre- and post-copulatory male harm mechanisms, and is perhaps the main model species in the study of sexual conflict (Arbuthnott et al., 2014; Chapman et al., 2003a; MacPherson et al., 2018; Malek and Long, 2019; Wigby and Chapman, 2005). During male-male pre-copulatory competition over access to mating, males harm females via intense harassment that causes physical injuries, interferes with female behaviors such as egg-laying or feeding, and results in energetically costly resistance to males (Bretman et al., 2019; Partridge and Fowler, 1990; Teseo et al., 2016). In the context of sperm competition over fertilizations, males can also harm females via toxic ejaculates, whereby certain male seminal fluid proteins manipulate female re-mating and egg-laying rates to the male’s advantage, but at a cost to females in terms of lifespan and lifetime reproductive success (Chapman et al., 2003b; Sirot et al., 2009; Wigby and Chapman, 2005). These proteins are secreted by male accessory glands, have been well characterized, and are strategically allocated by males in response to variation in the socio-sexual context (Hopkins et al., 2019; Sirot et al., 2011). Additionally, despite the fact that almost all work on male harm has, to date, been conducted in laboratory strains adapted to stable temperature conditions, D. melanogaster reproduces in the wild under a wide range of temperatures (Dukas, 2020; Kapun et al., 2018).

Briefly, we collected flies from a continental wild population in Requena (Spain) that experiences significant fluctuations in temperature even during the mildest months when it is reproductively active (e.g. July: average: 24.9°C, average min: 19.8°C, average max: 30.1°C, Fick and Hijmans, 2017). After acclimation of the resulting population to laboratory conditions under a fluctuating temperature regime mimicking natural conditions, we conducted five different experiments to gauge how temperature variation within a normal range (i.e. 20°C, 24°C, and 28°C) affects: (a) the overall impact of male-male competition on female lifetime reproductive success (i.e. male harm), (b) how the net effects of harm are accomplished in terms of different female fitness components (i.e. reproductive rate, actuarial aging, and reproductive aging), and (c) underlying male pre-copulatory (i.e. harassment) and post-copulatory (i.e. ejaculate toxicity) harm mechanisms (Figure 1).

Figure 1 with 5 supplements see all
Download asset Open asset
Schematic overview of the study.

(A) Our aim was to study how temperature variation, across a range at which reproduction is optimum in the wild, may affect: the net decrease in female fitness resulting from male harm, what female fitness components are mainly affected by male harm, and pre-copulatory (i.e. sexual harassment) and post-copulatory (i.e. ejaculate effects on female receptivity, short-term fecundity, and survival) mechanism of harm. (B) General design of the study: (1) We sampled a wild population of Drosophila melanogaster flies that reproduce optimally between 20°C and 28°C, (2) We setup a population in the lab and left it to acclimate for a few generations under a programmed fluctuating temperature regime that mimics wild conditions in late spring-early summer (20–28°C range with mean at 24°C), (3) We run a series of five experiments (each repeated at 20°C, 24°C, and 28°C) to study temperature effects on net male harm, female fitness components and male pre- and post-copulatory mechanisms of harm.

Materials and methods

Field collection

Request a detailed protocol

In October 2018, we used banana traps to sample Drosophila melanogaster flies from three wineries in Requena (Spain): ‘Hispano-Suizas’ (39.466128,–1.149642), ‘Pago de Tharsys’ (39.497834,–1.122781), and ‘Dominio de la Vega’ (39.515079,–1.143757). Traps were setup within the wineries, but in premises that were open to the exterior and to ample surrounding vineyards and/or in the vineyards themselves. After collection, we anesthetized flies using mild CO2 exposure. We then separated and individually distributed field-collected females in vials with standard food, left them to lay eggs for a period of 48 hr, and then incubated their eggs at 24°C, 60% humidity, and a 12:12 dark-light cycle for 14 days to allow adult flies to emerge. We inspected the genital arch of F1 males of each of these female isolines to distinguish D. melanogaster isofemale lines from D. simulans. We then collected three males and three females from each D. melanogaster isofemlae line (total of 276 flies from 46 isofemale lines) and released them into a population cage with a surplus of food medium supplemented with live yeast, setting up the ‘Vegalibre (VG)’ population. In November 2019 and October 2020, to maintain natural variation, we re-sampled the wineries and added 348 and 756 flies from 58 and 126 isofemale lines, respectively, following the same procedure (27 isofemale lines from ‘Pago de Tharsys’ and 31 isofemale lines from ‘Dominio de la Vega’ in November 2019 and 33 isofemale lines from ‘Pago de Tharsys,’ 50 isofemale lines from ‘Dominio de la Vega’ and 43 isofemale lines from ‘Hipano-Suizas’ in October 2020).

Stock maintenance and acclimation

Request a detailed protocol

We carried out all experiments between March 2020 and April 2021, using individuals from the VG field population kept in the laboratory with overlapping generations at an average temperature of 24°C with daily pre-programmed fluctuations (±4°C) mimicking natural daily temperature conditions during the reproductively active season, at ~60% humidity and on a 12:12 hr light:dark cycle (Pol Eko ST 1200 incubator). The lowest temperature was set up 1 hr after sunrise and the highest 1 hr after midday. It is important to note that our stock population of flies was kept under a programmed fluctuating temperature regime that mimics their average circadian rhythm in the field, but temperature fluctuations in nature will be inherently subject to minor stochastic variations whose effects we controlled for (and thus did not capture) in this experiment. We used maize-malt medium (7 g of agar, 72 g of malt, 72 g of maize, 9 g of soya, 15 g of live yeast, 20 g of molasses, and 33 ml of Nipagin mix –3 g of methyl 4-hydroxy-benzoate and 28 ml of ethanol 90%– per 1000 ml of water) as a food source throughout maintenance and experiments. To collect experimental flies, we introduced yeasted grape juice agar plates into stock populations to induce female oviposition. We then collected eggs and placed them in bottles containing ~75 ml of medium to be incubated at 24 ± 4°C at a mean density of 223 ± 14.3 (95% CI) (Clancy and Kennington, 2001). We collected virgin flies within 6 hr of eclosion, under ice anesthesia, and then sexed and kept them in vials with food until their use (~3 days later), at 24 ± 4°C (see below for more details).

Net impact of male harm on female fitness and underlying behavioural mechanisms (experiment 1)

Fitness assay

Request a detailed protocol

To study whether male harm is affected by temperature, we established a factorial design to measure survival and lifetime reproduction success (LRS) of female flies under monogamy (i.e. one male and one female per vial) vs. polyandry (i.e. three males and one female per vial), across three stable temperature treatments typical of this population during their reproductively active period in the wild: 20°C, 24°C, and 28 °C. Comparison of female fitness at monogamy vs. polyandry is standard procedure to gauge male harm in Drosophila and other organisms (see Yun et al., 2021 for a review and Gómez-Llano et al., 2023 for a recent meta-analysis). While the sex ratio in this species is typically 1:1, the operational sex ratio is male-biased and frequently reaches a 3:1 (or higher) male-bias in mating patches in the wild (Dukas, 2020). Thus, the 1:1 vs. 3:1 sex ratios used in this study represent biologically relevant scenarios and have actually become standard in Drosophila studies measuring male harm to females (Yun et al., 2021; Gómez-Llano et al., 2023).

We first collected virgin flies into same-sex vials of 15 individuals and then randomly allocated them to either of the three temperature treatments 48 hr before starting the experiment, at which temperatures they remained until its end. To estimate LRS, we transferred flies to fresh vials twice a week using mild CO2 anesthesia. We incubated the vials containing female eggs at 24 ± 4°C for 15–20 days (~15 days for vials coming from 28°C, ~17 days for 24°C and ~20 days for 20°C) to allow F1 offspring emergence, after which we froze them at –21°C for later counting. The differences in incubation time are due to differences in developmental time caused by temperature differences during the first 3–4 days of each vial (i.e. the time eggs remained at their respective temperature treatments before flipping females to new fresh vials and incubation at 24 ± 4°C). We discarded and replaced males with young (2–4 days old) virgin males (receiving the same treatment as described above for original males) three weeks after starting the experiment (at the same time for all treatments). In addition, we kept a stock of replacement males maintained at each of the three temperatures to replace dead male flies if needed. We kept focal female flies under these conditions for six weeks, after which we discarded males and followed females until they died for survival analysis (see Figure 1—figure supplement 1 for an overview of the experimental design).

We started the experiment with 468 females (78 per temperature and mating treatment) and 936 males (234 per temperature in polyandry and 78 per temperature in monogamy). Due to discarded (e.g. accidentally damaged during handling) and escaped flies, final (female) sample sizes were: (a) at 20°C, npolyandry = 74 and nmonogamy = 76, (b) at 24°C: npolyandry = 72 and nmonogamy = 77, and (c) at 28°C: npolyandry = 70 and nmonogamy = 75. We estimated the overall degree of male harm by calculating relative harm (H) following Yun et al., 2021:

H=Wmonogamy −WpolyandryWmonogamy 

where W corresponds to female fitness. Thus, H provides an estimate of the relative decrease in female fitness due to male harm.

Using the data collected above, we partitioned overall LRS effects into effects on early reproductive rate (i.e. offspring produced during the first two weeks of age), actuarial aging (i.e. lifespan), and reproductive aging (i.e. offspring produced over weeks 1–2 vs. 3–4). We used weeks 3–4 as an estimate of late reproductive rate because mortality was already evident at this point (i.e. reflecting aging) and then was very high from week 5 onwards (Figure 4—figure supplement 1; thus preventing accurate estimation of reproductive success).

Finally, we also calculated rate-sensitive fitness estimates for each individual female and treatment population. Rate-sensitive fitness estimates take into account when offspring are produced, not just how many offspring are produced, and thus allow estimating fitness subject to the population growth rate (Edward et al., 2011). It is important to understand how differences in the number and timing of offspring production translate into fitness under different demographic scenarios. For example, early reproduction is particularly favored in increasing populations whereas late reproduction gains in importance in decreasing populations. Thus, while LRS is most suited to estimate individual fitness in stable populations, rate-sensitive estimates are preferred when r ≠ 0 (Brommer et al., 2002). We calculated both individual (ωind) and population (ωpop) rate-sensitive fitness for the following intrinsic rates of population growth: r=–0.1, r=–0.05, r=0, r=0.05, and r=0.1 (see Edward et al., 2011 for a detailed account). We then used ωpop values to calculate the relative cost (Cr) of polyandry for each temperature treatment across different values of r as:

Cr=ωpop (polyandry)ωpop(monogamy)

Behavioral measures

Request a detailed protocol

Immediately after the fitness experiment started, we conducted behavioral observations on the first day of the experiment across all treatments (Figure 1—figure supplement 1). Our aim was to investigate the behavioral mechanisms that might underlie the potential fitness effects evaluated above. Due to logistic limitations, we conducted behavioral observations in the same temperature control room, so we had to conduct trials at 20°C, 24°C, and 28°C over three consecutive days (with both monogamy and polyandry treatments evaluated at the same time for each temperature), in a randomized order (i.e. 20°C, 28°C, and 24°C). Note that we collected virgin flies over three consecutive days to ensure all flies were 5 days-old at the start of the experiment. We recorded the following behaviors: (a) courtship intensity (number of courting males per female per hour), (b) male-male aggression rate (i.e. number of aggressions per hour), and (c) female rejection (i.e. number of rejections per hour; see Bastock and Manning, 1955; Connolly and Cook, 1973 for behavioral descriptions). We also recorded the number of total matings during the observation period.

Observations started at lights-on (10 a.m.) and lasted for 8 hr, during which time we continuously recorded reproductive behaviors using scan sampling of vials. Each complete scan lasted approximately 8 min, so that we always conducted one complete scan every 10 min to ensure the recording of all matings (see below). Scans consisted in observing all vials in succession for ca. 3 s each and recording all occurrences of the behaviors listed above (i.e. all-occurrences recording of target behaviors combined with scan sampling). We interspersed behavioral scans with very quick (<1 min) mating scans where we rapidly swept all vials for copulas at the beginning, in the middle, and at the end of each complete scan. This strategy ensured that we recorded all successful matings (>10 min), which typically last between 15 and 25 min in our population of D. melanogaster. We obtained a total of 49 scans per vial. Behavioral observations were conducted only once, on day 1 of the fitness experiment, as prior experiments have shown that courtship, aggressive and female rejection behaviors in D. melanogaster are sufficiently stable over time so that our behavioral indexes are representative of long-term treatment differences (e.g. Carazo et al., 2015; Carazo et al., 2014). In contrast to courtship, aggression, and rejection indexes, note that total mating frequency over the first day cannot be taken as a reliable measure of mating rate (Wolfner, 1997). Thus, our rationale in recording this variable was just to ensure that early mating ensued normally across treatments (which was the case, see Figure 5—figure supplement 2), but we did not include this variable in our statistical analyses.

Mating effect on female reproduction and survival (experiments 2 to 5)

To examine post-mating mechanisms that might underlie the fitness effects observed in our first experiment, we conducted four additional experiments to test whether temperature modulates the well-documented effects that mating with a male has on female receptivity, short-term fecundity, and survival. In D. melanogaster, males manipulate female reproduction via their ejaculate, which increases male fitness but frequently decreases female lifespan and lifetime reproductive success (Chapman et al., 1995). Briefly, males transfer seminal fluid proteins (SFPs) produced by their accessory glands that stimulate female short-term fecundity, decrease female receptivity, and ensure sperm storage, thus generally promoting male success in sperm competition (Chapman et al., 1995; Wigby and Chapman, 2005). In addition, prior studies have shown that males are able to tailor investment into SFPs according to perceived sperm competition risk (SCR) and intensity (Hopkins et al., 2019). Thus, we set-up a factorial design where we manipulated the temperature (i.e. 20°C, 24°C, and 28°C) and perceived SCR levels (i.e. males kept alone vs. with 7 more males in a vial) at which adult focal males were kept prior to mating.

Then, we measured how the reception of a treated male’s ejaculate after a single mating in a common garden environment (i.e. all matings at 24 °C) affected female fecundity, survival, and reproduction, following standard assays to gauge male ejaculate effects on females in D. melanogaster (e.g. Chapman et al., 1995; Perry et al., 2013; Wigby and Chapman, 2005; Wigby et al., 2009). We conducted separate experiments implementing two different temperature treatment durations (i.e. 48 hr and 13 days), to include two potential different scenarios. Our 48 hr treatment aimed to mimic short-term temperature effects on adult males whereas our 13 day treatment aimed to mimic longer-term effects on adult males that span a complete spermatogenesis cycle. The period from the synthesis of deoxyribonucleic acid in the spermatocyte to successful insemination is approximately 10 days long in D. melanogaster (Chandley and Bateman, 1962), so we treated males for 13 days after sperm depleting them (see below) to ensure that males experienced treatment temperatures across the whole spermatogenesis cycle.

Receptivity assays (experiments 2 and 3)

Request a detailed protocol

We first collected focal males as virgins (i.e. within 6 hr of eclosion) under ice anesthesia and randomly placed them either individually (low SCR) or in a same-sex group of eight (high SCR) in plastic vials with food. Next, we randomly divided them into three groups that we allocated to the different stable temperature treatments for either 48 hr (i.e. short treatment duration, experiment 2, Figure 1—figure supplement 2) or 13 days (i.e. long treatment duration, experiment 3, Figure 1—figure supplement 3) immediately before the beginning of each experiment. For experiment 3, we depleted the sperm and seminal fluid of focal males before allocating them to different temperature treatments by housing them with four standard virgin females for 24 hr, given that three successive matings are enough to deplete the accessory glands of male D. melanogaster (Linklater et al., 2007; Macartney et al., 2021).

We collected all females and competitor males (i.e. standard males without any previous treatment) used in receptivity assays as virgins and held them in same-sex groups of 15–20 flies at 24 ± 4°C. Experiments started by exposing all virgin females to single focal males for 2.5 hr at 24°C. After a successful copulation, we separated the mated females from the focal males and isolated them until the remating trial. We discarded unmated females and focal males. 72 hr after this first mating with the focal treated male, we individually exposed females to single virgin competitor males for 12 hr. After each trial, we transferred unmated females into a new vial with food, until the next remating trial on the next day (Figure 1—figure supplements 2 and 3). We repeated remating trials for three consecutive days, which allowed us to calculate the cumulative percentage of remated females (and associated re-mating latencies; see below) for each of the three days of each experiment. Due to a large number of vials/flies involved, we conducted the experiments in two blocks each: with n=390 females per batch in experiment 2 (n=436 rematings) and n=420 females per batch in experiment 3 (n=676 rematings). We also recorded mating duration for the first mating (i.e. with the focal treated male), the remating latency (i.e. the time lapse between males being introduced into the female-containing vial and copulation), and mating duration for re-matings. Females and focal and competitor males were 4 days old for experiment 2. In experiment 3, females and competitor males were 4 days old, while focal males were 18 days old.

Fecundity and survival assays (experiments 4 and 5)

Request a detailed protocol

To study the effects of a single mating on female short-term fecundity and long-term survival, we performed two experiments (experiments 4 and 5, Figure 1—figure supplements 4 and 5, respectively) where we compared female fecundity and F1 egg-to-adult viability of females mated with male flies subject to the same factorial design imposed in receptivity experiments (here experiment 4 had a treatment duration of 48 hr while experiment 5 had a treatment duration of 13 days). We collected and treated all focal males as in the receptivity assays described above, and then proceeded to mate virgin females in single pairs with focal males for 2.5 hr at 24°C. After copulation, we separated mated females from focal males and kept them individually in single vials. We discarded unmated females and focal males. We then transferred females to fresh vials every 24 hr for 4 days, and then every 3 days twice. Finally, we followed females until they died by combining them into same-treatment vials of 10 females that were flipped once a week. We removed dead flies at each flip and scored deaths on a daily basis. We counted eggs laid during the first 3 days and incubated vials from days 1, 2, 3, 4, 5, and 8 until adults emerged to count progeny and determine egg-to-adult viability (Figure 1—figure supplements 4 and 5). Sample sizes were 545 females for experiment 4, and 480 females for experiment 5. Females and focal or competitor males were 4 days old for experiment 4. In experiment 5, females and competitor males were 4 days old, while focal males were 18 days old.

Statistical analyses

We performed all statistical analyses using R statistical software (version 3.5.2). In all cases, we assessed fit and validated models by visual inspections of diagnostic plots on raw and residual data (Zuur et al., 2010). In all models, we used ANOVA type III test ‘F’ to compute p-values corrected by the Benjamini-Hochberg (BH) method to control the inflation of the type I error-rate due to multiple testing. We fitted all models with the temperature effect as a covariate, given that it is a continuous variable. In all cases where we detected a significant interaction between main effects, we ran models separately for each temperature (or treatment duration for experiments 2–5) to explore the nature of such interactions. As a complementary analysis, in these cases, we re-fitted the original model with temperature as a factor, which allowed us to run a post hoc Tukey’s test as an additional way to explore interactions while controlling for inflation of experiment-wise type 1 error rate. We provide the latter in the SM, but we note that both approaches always yielded qualitative identical results.

Experiment 1

Request a detailed protocol

To examine temperature effects on male harm, we evaluated the interaction between mating system and temperature on female fitness (LRS), early reproductive rate, reproductive aging, actuarial aging, and male and female reproductive behaviors (courtship intensity and female rejection; experiment 1). We fitted generalized linear models (GLMs) with temperature, mating system, and their interaction as fixed effects. Graphical inspection of LRS, actuarial aging, and reproductive behaviors (courtship intensity and male-male aggression) revealed that the normality assumption was apparently violated, as well as the independence assumption for LRS. Box–Cox transformation (Quinn and Keough, 2002) solved these problems and allowed us to run a GLM with a Gaussian error distribution. We compared GLMs with their corresponding null GLMs using the likelihood ratio test only to test the significance of the independent variables in the full model. We detected collinearity between the mating system and the interaction in LRS, early reproductive, reproductive aging, actuarial aging, courtship intensity, and female rejection models. In all these cases, we thus refitted the model without the main mating system effect (which was not our main interest). As a complementary analysis for LRS, we also ran a model with temperature as a factor and a predetermined quadratic contrast table (given the relationship between LRS and temperature is clearly non-linear, Figure 2), and obtained similar results.

Figure 2 with 2 supplements see all
Download asset Open asset
Female lifetime reproductive success (mean ± SEM) across temperature and mating system treatments.

20°C: npolyandry = 73 and nmonogamy = 74. 24°C: npolyandry = 71 and nmonogamy = 74. 28°C: npolyandry = 66 and nmonogamy = 71.

We also used Cox proportional hazards survival model to analyze potential differences in mortality risk across treatments, using the survival and survminer packages (Therneau, 2015; Kassambara and Kosinski, 2018), which allowed us to include the females lost during manipulations as ‘right censored’ individuals (i.e. individuals that are taken into account for demographic analysis until the day they disappear, Kleinbaum and Klein, 2012). We analyzed female rejection behaviors in two different ways. First, we examined mating system and temperature effects on overall rejection rates using a Gamma distribution, as this variable was continuous and zero-inflated. Second, we also used a binomial GLM to estimate potential differences in female rejections per courtship. Finally, we analyzed male-male aggression with the mating system as the sole main factor (as male-male aggressions where only possible in the polyandry treatment).

Experiments 2–5

Request a detailed protocol

To examine the effect of temperature on post-copulatory effects, we evaluated the effect of SCR, temperature, treatment duration, and their interaction on receptivity (mating duration and remating latency -experiments 2 and 3) and fecundity -oviposition and egg viability- and female survival (experiments 4 and 5). For mating duration, remating latency and egg viability we fitted generalized linear models (GLMs) with temperature, SCR level, treatment duration, and their interaction as fixed effects. We assessed the significance of factors by dropping individual terms from the full model using the ‘drop1’ function, refitting models without the triple interaction where necessary. We detected a problem of collinearity between SCR and the interactions, as well as between treatment duration and the interactions in mating duration, remating latency, and egg viability models. In all these cases, we refitted the model without the main SCR and treatment duration effects (which were not our main interest). For the mating duration, we used a Gamma distribution. For remating latency and egg fertility, we used a Gaussian distribution. For oviposition, we fitted a generalized linear mixed model (GLMM) with temperature, SCR level, treatment duration and their interaction as fixed effects and day as a random effect. Initially, we run a model with a zero-inflated distribution, in which the zero values are modeled separately from the non-zero values (Zuur et al., 2010). However, we detected problems of collinearity, including treatment duration as an effect. We thus run two separate models for each treatment duration using Hurdle models without the main SCR effect. Finally, for survival, we used a Cox proportional hazards survival model to analyze potential differences in mortality risk across treatments, including the females lost during manipulations as ‘right censored’ individuals.

Results

Net impact of male harm on female fitness and underlying behavioral mechanisms (experiment 1)

LRS

We detected a significant temperature by mating system interaction for female lifetime reproductive success (F1,425 = 16.931, p<0.001, Figure 2, Figure 2—figure supplement 1), and a marginally non-significant temperature effect (F1,425 = 3.712; p=0.054). Separate models for each temperature level show a larger effect of the mating system on lifetime reproductive success at 24°C than at 20°C, and larger at 20°C than at 28°C, despite 95 % CI of the estimates overlaps (Table 1, Table 1—source data 1a). The decrease in LRS in polyandry vs. monogamy peaked at 24°C (H=0.36) and was 1.6 (H=0.22) and 3.4 times (H=0.10) smaller at 20°C and 28°C, respectively. Rate-sensitive fitness estimates show that estimated population costs are dependent on background growth rates (Figure 3), and in general particularly accused in decreasing populations.

Table 1
Output from separate generalized linear models (GLMs) for each temperature level to explore significant interactions between temperature and mating system effects on female fitness components.
T°CLRSReproductive agingActuarial aging
Fdfp-valueEstimate (95% CI)Fdfp-valueEstimate (95% CI)Fdfp-valueEstimate (95% CI)
20°4.41,1450.0391.07
(0.06–2.06)		12.11,145<0.001–7.99
(−12.5--3.5)		39.6 1,148<0.0017.44
(5.1–9.8)
24°16.61,142<0.00122.39
(11.6–33.1)		35.31.142<0.001–17.2
(−22.9- -11.5)		32.2 1,143<0.0014.84
(3.2–6.5)
28°2.21,1350.1371.88
(−0.58–4.36)		14.11,135<0.001–11.87
(−18.1- -5.7)		19.7 1,137<0.0012.97
(1.7–4.3)
Table 1—source data 1

Summary statistics from Tukey’s post hoc test to examine the meaning of significant interactions between temperature and mating system effects.

(a) Polyandry – Monogamy contrast table for each temperature level for female fitness components. (b) Polyandry – Monogamy contrast table for each temperature level for underlying behavioral mechanisms. Test from generalized linear models (GLMs) fitted with temperature as factor. Note that using Tukey’s post hoc yielded qualitatively identical results from running models separately for each temperature.

https://cdn.elifesciences.org/articles/84759/elife-84759-table1-data1-v2.docx
Table 1—source data 2

Summary statistics from Cox PH survival models as a complementary analysis to examine potential differences in mortality risk across treatments from the experiment 1.

(a) Summary statistics from Cox PH survival full model. (b) Summary statistics from fitting separate Cox PH models for each temperature level due to a significant interaction between temperature and mating system. (c) Polyandry – Monogamy contrast table from Tukey’s post hoc for each temperature level from Cox PH survival model fitted with temperature as factor. p-values from Cox HP models are computed using ANOVA type III, LR test. Note that using Tukey’s post hoc yielded qualitatively identical results from running models separately for each temperature. The corresponding survival plot is plotted in Figure 4—figure supplement 1 .

https://cdn.elifesciences.org/articles/84759/elife-84759-table1-data2-v2.docx
Figure 3
Download asset Open asset
Rate-sensitive fitness estimates.

(a) Average rate-sensitive index fitness estimate of individual females (Mean ωind) for different population growth rates across temperature and mating system treatments (shaded areas denote SEM). 20°C: npolyandry = 73 and nmonogamy = 74. 24°C: npolyandry = 71 and nmonogamy = 74. 28°C: npolyandry = 66 and nmonogamy = 71. (b) Relative cost (Cr) of polyandry (vs. monogamy) for each temperature treatment for different population growth rates. Cr was calculated based on rate-sensitive index fitness estimates for populations (ωpop), whereby population costs are shown as 1 – Cr, thus reflecting the relative decrease in population growth rate.

Early reproductive rate

We did not detect a significant temperature by mating system interaction (F1,425 = 2.94; p=0.09). We did detect a significant main temperature effect (F1,425 = 64.63; p<0.001; Figure 2—figure supplement 2), such that early reproduction increased at 28°C.

Reproductive aging

We detected a significant temperature by mating system interaction for reproductive aging (F1,425 = 55.24; p<0.001; Figure 2—figure supplement 2), and a clear main effect for temperature (F1,425 = 44.25; p<0.001). Running models separately for each temperature level showed that mating system affected reproductive aging at all temperatures, but particularly so at 24°C and 28°C (Table 1, Table 1—source data 1a, Figure 2—figure supplement 2).

Actuarial aging

We detected a significant temperature by mating system effect for lifespan (F1,428 = 73.81; p<0.001; Figure 4, Figure 4—figure supplement 1a), and a significant main effect for temperature (F1,428 = 36.98; p<0.001). Mating system affected actuarial aging at all temperatures, but particularly so at 20°C (Table 1, Table 1—source data 1). The survival analysis yielded qualitatively identical results (Table 1—source data 2, survival plot Figure 4—figure supplement 1b).

Figure 4 with 1 supplement see all
Download asset Open asset
Male harm effect on female lifespan (mean ± SEM) across temperature and mating system treatments.

20°C: npolyandry = 73 and nmonogamy = 74. 24°C: npolyandry = 71 and nmonogamy = 73. 28°C: npolyandry = 66 and nmonogamy = 73.

Reproductive behaviour

The interaction between temperature and mating system was significant for courtship rate (F1,441 = 45.62; p<0.001), and we also detected a main temperature effect (F1,441 = 16.69; p<0.001 Figure 5a, Figure 5—figure supplement 1a). Running models separately for each temperature level, mating system affected courtship rate at 24°C and 28°C but not at 20°C (Table 2, Table 1—source data 2b). Likewise, we detected a significant temperature by mating system effect for rejection rate (F1,441 = 24.48; p<0.001 Figure 5b, Figure 5—figure supplement 1b), and a main effect for temperature (F1,441 = 5.61; p=0.020). Models for each temperature level show a mating system effect at 24°C and 28°C but not at 20°C (Table 2, Table 1—source data 2b). We did not detect a significant interaction between temperature and mating system (F1,294 = 0.30; p=0.582), nor a temperature effect (F1,294=0.08; p=0.773), for rejection rates per courtship. For male-male aggression rate, we detected a significant temperature effect (F1,214 = 14.45; p<0.001 Figure 5c, Figure 5—figure supplement 1c).

Figure 5 with 2 supplements see all
Download asset Open asset
Reproductive behaviors (mean ± SEM) across temperature and mating system treatments.

(a) Courtships per female per hour, (b) Female rejections per hour, and (c) Aggressions male-male per hour. 20°C: npolyandry = 74 and nmonogamy = 76. 24°C: npolyandry = 72 and nmonogamy = 77. 28°C: npolyandry = 70 and nmonogamy = 75.

Table 2
Output from separate generalized linear models (GLMs) for each temperature level to explore significant interactions between temperature and mating system effects on underlying behaviorual mechanisms.

p-values were corrected for multiple testing using Benjamini-Hochberg correction.

T°CCourtship rateRejection rate
Fdfp-valueEstimate (95% CI)Fdfp-valueEstimate (95% CI)
20°0.41,1480.546–0.04
(−0.16–0.08)		0.201,1480.654–0.05
(−0.30–0.19)
24°21.81,147<0.001–0.40
(−0.57- -0.23)		10.91.1470.001–17.2
(−1.01- -0.25)
28°40.21,143<0.001–0.63
(−0.83- -0.43)		19.31,143<0.001–11.87
(−0.96- -0.36)

Mating effects on female reproduction and survival (experiments 2 to 5)

Female receptivity (experiments 2 and 3)

For the duration of the first mating in our female receptivity assays, we detected significant SCR by temperature (F1,1239 = 40.42; p<0.001), treatment duration by temperature (F1,1239 = 5.97; p=0.024), and SCR by treatment duration (F1,1239 = 5.48; p=0.024) interactions (Figure 6a, Figure 6—figure supplement 1a). We found no significant main effect for temperature (F1,1239 = 3.36; p=0.07). Running models separately for each temperature, SCR affected the duration of the first mating at 20°C, 24°C, and 28°C, while treatment duration only affected the duration of the first mating at 24°C (Table 3a, Table 3—source data 1a-b). Additionally, running models for each treatment duration, SCR affected the duration of the first mating at short and long treatment durations (Table 3b, Table 3—source data 1c). Similarly, for female remating latency, we detected significant interactions for SCR by temperature (F1,1094 = 6.15; p=0.022), and treatment duration by temperature (F1,1094 = 5.17; p=0.028), whereas the interaction between SCR by treatment duration was not significant (F1,1094 = 1.00; p=0.316) (Figure 6b, Figure 6—figure supplement 1b). We also detected a significant main temperature effect (F1,1094 = 8.21; p=0.01). Running models separately for each temperature level, SCR level, and treatment duration only affected remating latency at 28°C (Table 3a, Table 3—source data 1a-b).

Figure 6 with 4 supplements see all
Download asset Open asset
Mean ± SEM for mating duration and remating latency.

(a) Mating duration of males exposed to high (8 males per vial) or low sperm competition risk (1 male per vial) for 48 hr or 13 days prior to mating at different temperatures. 20°C: nhigh/48hr = 91, nlow/48hr = 96, nhigh/13days = 121 and nlow/13days = 117. 24°C: nhigh/48hr = 85, nlow/48hr = 88, nhigh/13days = 119 and nlow/13days = 115. 28°C: nhigh/48hr = 92, nlow/48hr = 104, nhigh/13days = 99 and nlow/13days = 117. (b) Female remating latency following a single mating with either a male from a high or low sperm competition risk level for 48 hr or 13 days before mating across temperature treatments. 20°C: nhigh/48hr = 75, nlow/48hr = 73, nhigh/13days = 119 and nlow/13days = 113. 24°C: nhigh/48hr = 61, nlow/48hr = 70, nhigh/13days = 116 and nlow/13days = 113. 28°C: nhigh/48hr = 63, nlow/48hr = 82, nhigh/13days = 98 and nlow/13days = 117.

Table 3
Model outputs from separate generalized linear models (GLMs) for each (a) temperature level and (b) treatment duration to explore significant interactions.
a)
T°CEffectMating durationRemating latency
Fdfp-valueEstimate (95% CI)Fdfp-valueEstimate (95% CI)
20°Sperm competition risk3.91,4230.0460.03
(0.0005- 0.05)		0.951,3770.33027.9
(−28.2– 84.2)
Treatment duration2.31,4230.133–0.02
(−0.04- 0.006)		0.00061,3770.980–0.73
(−58.5– 57.0)
24°Sperm competition risk10.61,4050.0010.05
(0.02– 0.07)		0.071,3580.779–8.47
(−67.7– 50.7)
Treatment duration3.71,4050.054–0.02
(−0.05– 0.0003)		0.041,3580.842–6.24
(−67.8– 55.3)
28°Sperm competition risk26.51,410<0.0010.084
(0.052– 0.117)		8.051,3580.00587.81
(27.1– 148.4)
Treatment duration0.61,4100.451–0.12
(−0.04- -0.2)		9.731,3580.002–97.65
(−158.9– -36.3)
b)
TreatmentdurationMating duration
Fdfp-valueEstimate (95% CI)
Short (48 hr)4.51,5540.0330.03
(0.002– 0.06)
Long (13 days)54.21,686<0.0010.07
(0.051– 0.089)
Table 3—source data 1

Summary statistics from Tukey’s post hoc test as a complementary analysis to examine the meaning of significant interactions found for mating duration and remating latency.

(a) High – low sperm competition risk contrast table for each temperature level. (b) Long – short treatment duration contrast table for each temperature level. (c) High – low sperm competition risk contrast table for each treatment duration. Test from generalized linear models (GLMs) fitted with temperature as factor. Note that using Tukey’s post hoc yielded qualitatively identical results from running models separately for each temperature or treatment duration.

https://cdn.elifesciences.org/articles/84759/elife-84759-table3-data1-v2.docx

Female fecundity and survival (experiments 4 and 5)

For the number of eggs produced by females during the three first days, we did not detect significant interactions between temperature and SCR for either short or long treatment durations, nor a main significant effect for temperature (Figure 6—figure supplement 2 & Table 4—source data 1): (i) Short treatment duration, SCR by temperature interaction (χ21=0.05; p=0.821), temperature effect (χ21=2.82; p=0.092), (ii) Long treatment duration, SCR by temperature interaction (χ21=0.03; p=0.840), temperature effect (χ21=0.37; p=0.541). Likewise, for the total number of offspring produced by females during days 1, 2, 3, 4, 5, and 8 after mating, we did not find significant interactions between SCR and treatment duration (F1,952 = 0.022; p=0.881), or between SCR and temperature (F1,952 = 0.418; p=0.674), but we did between temperature and treatment duration (F1,952 = 9.599; p=0.005) (Figure 6—figure supplement 3). We did not detect a main temperature effect (F1,952 = 2.797; p=0.157). Running models for each temperature level, we found that treatment duration affected the total number of offspring produced by females at 24°C and 28°C, but not at 20°C (Table 4, Table 4—source data 2).

Table 4
Summary statistics from fitting generalized linear models (GLMs) separately for each temperature level to explore the significant interaction between temperature and treatment duration effects for total offspring produced by females during days 1, 2, 3, 4, 5, and 8 after mating.
T°CTotal of offspring
Fdfp-valueEstimate (95% CI)
20°0.61,3220.4541.42 (−2.30–5.15)
24°4.61,3210.0324.11 (0.35–7.86)
28°5.21,3080.0224.26 (0.62–7.89)
Table 4—source data 1

Summary statistics from the Hurdle model to analyze potential differences in egg production across treatments with temperature as a factor.

Note that using temperature as a factor yielded qualitatively identical results than treating it as a continuous covariable. p-values from Hurdel model are computed using ANOVA type III, Wald test. Corresponding data is plotted in Figure 6—figure supplement 2.

https://cdn.elifesciences.org/articles/84759/elife-84759-table4-data1-v2.docx
Table 4—source data 2

Summary statistics from Tukey’s post hoc test as a complementary analysis to examine the meaning of significant interaction between temperature and treatment duration for total of offspring produced by females during the days 1, 2, 3, 4, 5, and 8 after mating.

Short (48 hr) – Long (13 days) treatment duration contrast table for each temperature level. Test from generalized linear models (GLMs) fitted with temperature as factor. Note that using Tukey’s post hoc yielded qualitatively identical results from running models separately for each temperature.

https://cdn.elifesciences.org/articles/84759/elife-84759-table4-data2-v2.docx

Finally, we did not detect significant interactions in survival (SCR by treatment duration, χ21=0.276, p=0.694; SCR by temperature, χ21=0.311, p=0.694; temperature by treatment duration, χ21=4.09, p=0.128), nor significant main temperature (χ21=1.69; p=0.386) or SCR effects (χ21=0.154; p=0.694). We did detect a significant main treatment duration effect (χ21=13.03; p=0.001; Figure 6—figure supplement 4).

Discussion

We show that male harm exhibits complex plasticity in response to temperature changes within an optimal reproductive range (20–28°C) for a wild D. melanogaster population, with several implications for our understanding of how sexual conflict unfolds in nature. First, we show that net harm to females varies markedly within this range, such that relative harm (i.e. the proportional reduction in average female LRS in polygamy vs. monogamy) was maximal at 24°C, decreased at 20°C and was minimum at 28°C. Second, rate-sensitive fitness estimates indicate that arising population costs are dependent on the interaction between temperature and population demography, such that demography modulates the impact of male harm on population viability less at warmer temperatures. Third, our results strongly suggest that male harm effects on population growth have to do with the fact that different mechanisms exhibit qualitatively different reaction norms in response to temperature, with distinct effects on different female fitness components. More specifically, at cold temperatures courtship intensity (i.e. male harassment; pre-copulatory harm mechanism) decreased, and female fitness was impacted more via accelerated actuarial aging than at warm temperatures. In contrast, warm temperatures impacted mating costs and effects on female receptivity (i.e. post-copulatory mechanism), and female fitness decreased more via accelerated reproductive aging than at colder temperatures. We discuss how such plasticity may affect how male harm impacts populations, as well as selection, adaptation and, ultimately, evolutionary rescue under a warming climate.

Temperature effects on male harm and its consequences for populations

We found that temperature variation within the optimal reproductive range for our study population in the wild had a strong effect on net male harm levels. To gauge male harm, we used the standard procedure of comparing female LRS in monogamy, which imposes low male-male competition and thus low sexual conflict, vs. polyandry (i.e. a female with three males), which imposes high male-male competition and intensifies sexual conflict between the sexes (Yun et al., 2021). These sex ratios are common in mating patches in the wild, and are actually representative of the extremes in natural levels of male-male competition (Dukas, 2020). In monogamy, temperature did not affect female fitness (Figure 2), showing that female reproduction is indeed optimal within this range. In contrast, the net decrease in female LRS in polyandry vs. monogamy was highly dependent on the thermal environment, with an average decrease of H=0.36 at 24 °C, H=0.22 at 20 °C. and H=0.10 at 28 °C, at which temperature we did not find a statistically significant effect of mating system on female LRS (Figure 2).

Male harm effects are expected to be cumulative over time, so that their impact on female survival and reproductive output is unlikely to be constant across a female’s lifespan (Bonduriansky et al., 2008; Filice et al., 2020). At the same time, early vs. late-life reproduction weigh differently on both individual fitness and how this impacts background population growth, depending on whether such population is decreasing, stable, or growing (Edward et al., 2011; Priest et al., 2008). Thus, in order to evaluate how male harm is likely to impact populations across different temperatures, we calculated rate-sensitive fitness estimates for individual (Figure 3a) and population (Figure 3b) fitness across a range of demographic scenarios (i.e. decreasing, stable, and growing populations; Edward et al., 2011). Overall, the impact of male harm was higher in decreasing populations, where late-life reproduction gains importance, which is consistent with the idea that male harm effects are cumulative over the lifespan of females. Above and beyond this general effect, the observed interaction between temperature and female fitness is maintained irrespective of population demography. That is, male harm decreases female individual fitness more at 24°C than at 20°C and 28°C. Interestingly, though, temperature also has a clear effect on how demography affects population-level costs. At hotter temperatures the relative population costs of male harm vary considerably less with demography (i.e. background population growth) than at colder temperatures (Figure 3b). For a decreasing population, the relative population costs of male harm are significantly higher at 20°C than at 28°C, but this difference wanes as population growth rate (r) increases, to the point of reverting in a rapidly growing population (Figure 3b). This suggests that male harm has more impact on late-life female fitness at cold temperatures, and hints at the possibility that cold vs. hot temperature affect qualitatively distinct parameters of female fitness, and thus underlying mechanisms of male harm.

Looking at the effects on separate female fitness components yields results largely in agreement with the above ideas. Again, consistent with the fact that harm needs to accumulate in time to impact female fitness, temperature had no effect on how or whether male competition impacted early reproductive rate (Figure 2—figure supplement 2). Temperature did, however, modulate how male harm impacted on female actuarial vs. reproductive aging. We found clear differences in female lifespan across temperature and mating system treatments. Male harm effects on actuarial aging (an increase in mortality rate with age) were more severe at 20°C (35% decrease in female lifespan) than at 24°C (31% decrease) and at 28°C (22% decrease; Figure 3). In contrast, while male harm accelerated reproductive aging at all temperatures, this decrease was more marked at 24°C and at 28°C than at 20°C (Table 1; Figure 2—figure supplement 2). In accordance with available evidence in lab-adapted flies (García‐Roa et al., 2019), these results show that temperature does not seem to have a linear effect on aging processes.

To sum up, we offer strong evidence that different male harm mechanisms are sensitive to temperature in different ways, with ensuing modulation of its effects on different female fitness components. We underscore two potential consequences arising from these findings. The first is that the net fitness effects of male harm might be lower than expected when considered in its natural thermal setting because previous research has focused on studying male harm at average temperatures, precisely where we found it to be maximal. It follows that integrating harm across the natural temperature range will result in a lower net decrease in female fitness in the wild. The second is that environmental variability may foster the maintenance of genetic variation underlying different mechanisms of male harm, and even potentially divergent male-male competition strategies. Our work joins an increasing number of recent studies in highlighting the importance of evaluating more ecologically realistic scenarios in sexual conflict research, particularly how natural fluctuations in the ecology of the socio-sexual context may affect sexual conflict processes (García‐Roa et al., 2019, García-Roa et al., 2020; Gomez-Llano et al., 2018; Yun et al., 2017).

In addition, the above results open up the possibility that warm climates may lessen the impact of sexual conflict on population viability, perhaps facilitating evolutionary rescue. Male harm effects were found to be relatively lower in warmer temperatures and in decreasing populations, precisely the type of context that would be typical of a climate-change scenario. The effects found in this study were within the optimum reproductive range for this population, but similar results have been reported in response to stressful temperatures. For instance, temperature has been used to induce environmental stress in natural populations of seed beetles (Callosobruchus maculatus), showing that a stressful thermal regime reduces intra-locus sexual conflict by aligning selection in males and females (Berger et al., 2014, but see Martinossi-Allibert et al., 2019). A previous study in a lab-adapted population of D. melanogaster also shows that male harm levels (i.e. inter-sexual conflict) decrease when subject to maladaptive warm temperatures (García‐Roa et al., 2019). However, there are two reasons why it is important to caution against direct extrapolation of our results to wild populations at this stage, in particular in relation to their relevance for populations facing the current climate crisis. First, our monogamy vs. polyandry treatments reflect the low vs. high-end of the spectrum of operational sex ratios that are typical of D. melanogaster at mating patches in the wild (Dukas, 2020), and thus our measures of male harm effects are likely to be generally higher than expected in nature. While this does not change the main conclusions regarding temperature effects, it is important to note when considering the degree to which these effects may be relevant in the wild. Second, our treatment temperatures were stable, designed to study how coarse-grain changes in temperature across the adult lifespan of flies may influence how sexual conflict unfolds in nature. Thus, future studies will need to encompass how fine-grained fluctuation (i.e. repeated variation of temperature across an adult’s lifespan) may affect male harm for a more comprehensive picture of temperature effects on sexual conflict in the wild.

Temperature effects on sexual conflict mechanisms in Drosophila melanogaster

Prior to mating, D. melanogaster males harm females via sexual harassment, due to protracted courtship of one or several males that results in physical damage, interference with other behaviors (e.g. foraging or egg-laying), and costly energetic investment into male avoidance (e.g. female rejection) (Bretman et al., 2019; Partridge and Fowler, 1990; Teseo et al., 2016). Importantly, previous studies in this species have shown that male harm is directly related to courtship intensity and female rejection, and indirectly to male-male aggression as a direct measure of male intrasexual competition (e.g. Bretman et al., 2019; Carazo et al., 2014; Partridge and Fowler, 1990). In our study, we found a clear increase in both courtship intensity and female rejection in polyandry vs. monogamy, but this was largely dependent on the thermal environment (Figure 5a–b). While we found clear evidence that harassment increases in polygamy at 28°C and 24°C, we did not find a similar effect at 20°C (Figure 5a). Female rejection behaviors exhibited the same trend (Figure 5b), and this was due to increased male courtship attempts (not to an increase in female likelihood to reject male courtships). Thus, our results suggest that male harassment decreases drastically at cold temperatures and is perhaps maximal at warm temperatures, at which temperatures we also detected the highest level of male-male aggression (Figure 5c). The above results seem to suggest that the decrease in male harm to females that we detected at 20°C vs. 24°C (Figure 2) is likely explained by the substantial drop in male harassment at this temperature. However, the same logic cannot apply to the decrease in male harm to females that we detected at 28°C vs. 24°C (see below).

During mating, D. melanogaster males transfer seminal fluid proteins (SFPs) that manipulate female re-mating and egg-laying rates to their advantage, but this normally comes at a cost to females in terms of lifespan and lifetime reproductive success (Chapman et al., 2003b; Chapman et al., 1995; Sirot et al., 2009; Wigby and Chapman, 2005). Furthermore, these effects are known to be modulated by the socio-sexual context, such that males strategically adjust their investment in SFPs depending on expected SCR levels (Bretman et al., 2009; Hopkins et al., 2019; Wigby et al., 2009). We run a series of standard assays in Drosophila (Bretman et al., 2009; Hopkins et al., 2019; Wigby et al., 2009) to investigate whether temperature modulates any of the known phenotypic effects that SFPs have on females following a single mating with males that were exposed to low (i.e. kept alone in their vial) or high SCR (i.e. kept with seven other males in a vial). The temperature males were kept at prior to mating did not modulate how mating affected short-term female fecundity or lifespan, but we did detect a clear effect of temperature with respect to female receptivity. Generally, reception of a male ejaculate resulted in a sharper decrease in female receptivity (i.e. longer remating latency) when males perceived a high SCR, which is to be expected and is in accordance with previous studies (Bretman et al., 2009; Bretman et al., 2010; Denis et al., 2017). However, this effect was largely modulated by temperature, such that it was more clearly detected at 28°C (Figure 6b). Interestingly, this effect was consistent for males treated for 48 hr and 13 days (albeit more marked in the latter case), and these results were paralleled by temperature effects on mating duration (Figure 6a). Namely, males exposed to high SCR generally mated for longer than males exposed to low SCR, again in line with previous findings (Bretman et al., 2010; Bretman et al., 2013), but this difference was clearly larger at 28°C (Figure 6a). It is worth restating that, under monogamous conditions, we did not detect a decrease in female productivity at 28°C vs. 24°C and 20°C, so that general differences in sperm viability are unlikely to account for the aforementioned differences. A potential explanation for these results is that males may perceive a higher SCR when kept in groups at warmer temperatures (e.g. due to increased activity). This, however, would predict the same differences between 24°C and 20°C, which was not the case.

All in all, our results suggest that at least some post-copulatory harm mechanisms are sensitive to temperature, because receipt of a male ejaculate resulted in a sharper decrease of female receptivity in high vs. low SCR at warm temperatures (particularly after 13 days of exposure). We speculate that this may contribute to explain why male harm drops so sharply at 28°C despite the fact that male harassment and male-male competition seem to be maximal at this temperature. In D. melanogaster, as in many other species, repeatedly mating is costly for females in terms of lifetime reproductive success (Arnqvist and Nilsson, 2000; Chapman et al., 1995; Fowler and Partridge, 1989). An intriguing possibility is thus that either SFPs are more effective at lowering female re-mating rates or males are investing relatively more in SFPs with increasing SCR at warm temperatures, thereby buffering these costs. An alternative (but complementary) possibility is that temperature may affect female behaviour or physiology in a way that makes them more resistant to harm.

We suggest future studies should explore these ideas by examining in detail how temperature affects the composition and transfer of SFPs to females, and how females respond to the transfer of these proteins and to male harm in general (i.e. effects on female resistance). In combination with experimental evolution at different temperatures, such an approach would allow us to disentangle between two causal hypotheses for the observed results. First, that warm temperatures may buffer sexual conflict in itself by aligning male and female reproductive interests. For example, if live-fast-die-young strategies fare relatively better for females at warm than cold temperatures, male and female optimal reproductive strategies may overlap more due to the fact that cumulative late-life effects of male harm might be diluted by the inherently high female intrinsic mortality at warm temperatures. Second, whether modulation of male harm at cold and (particularly) warm temperatures has to do with the fact that different male harm mechanisms are adapted to operate better at certain temperatures. For example, due to environmental effects on male activity or protein folding. In the latter case, male harm would be expected to increase as males adapt to higher or lower average temperatures, but sexual conflict per se (i.e. the degree to which male and female evolutionary interests overlap) would be expected to remain constant. Both of the above hypotheses could have broad consequences for our understanding of the evolution of sexual conflict across the tree of life.

Conclusions

Our findings may have implications for our understanding of how sexual conflict unfolds in nature, and its consequences for populations. First, they add to growing evidence (Gomez-Llano et al., 2018; MacPherson et al., 2018; Malek and Long, 2019; Perry and Rowe, 2018; Yun et al., 2017) indicating that ecological context is key in shaping sexually antagonistic coevolution and, in particular, suggest that temperature may be a particularly salient ecological factor to understand how sexual conflict evolves and operates in nature (García-Roa et al., 2020). Second, they highlight that male harm mechanisms can be highly plastic even in response to relatively minor fluctuations in temperature well within the optimal reproductive range, and suggest that different harm mechanisms are differently affected by temperature. Third, they suggest that male harm effects on female life-history and fitness components are asymmetrically modulated by temperature; male harm particularly decreased survival at cold and moderate temperatures, and reproductive aging at moderate and hot temperatures. In conjunction, these phenomena may have a bearing on evolutionary rescue and local adaptation processes. For example, in maintaining genetic variation in sexually selected traits in males, and/or in ameliorating the demographic impact of sexual conflict in populations facing environmental change.

Data availability

All data generated or analysed during this study are included in the manuscript. Source data files are uploaded to Dryad repository (https://doi.org/10.5061/dryad.pzgmsbcqz), along with R script https://doi.org/10.5281/zenodo.7350587.

The following data sets were generated
    1. Londoñ-Nieto C
    2. García-Roa R
    3. Garcia-Co C
    4. González P
    5. Carazo P
    (2023) Dryad Digital Repository
    Thermal phenotypic plasticity of pre- and post-copulatory male harm buffers sexual conflict in wild Drosophila melanogaster.
    https://doi.org/10.5061/dryad.pzgmsbcqz

References

    1. Clancy DJ
    2. Kennington J
    (2001)
    A simple method to achieve consistent larval density in bottle cultures
    Drosophila Information Service 84:168–169.
  46. Software
    1. Kassambara A
    2. Kosinski M
    (2018)
    Survminer: drawing survival curves using "ggplot2"
    Version 0.4.3, R Package.
    1. Kokko H
    2. Brooks R
    (2003)
    Sexy to die for? sexual selection and the risk of extinction
    Annales Zoologici Fennici 40:207–219.
  62. Book
    1. Parker GA
    (1979) Sexual selection and sexual conflict
    In: Blum M, Blum NA, editors. Sexual Selection and Reproductive Competition in Insects. Academic Press. pp. 123–166.
    https://doi.org/10.1016/B978-0-12-108750-0.50010-0
    1. Perry JC
    2. Rowe L
    (2018) Sexual conflict in its ecological setting
    Philosophical Transactions of the Royal Society B 373:20170418.
    https://doi.org/10.1098/rstb.2017.0418
    1. Plesnar-Bielak A
    2. Łukasiewicz A
    (2021) Sexual conflict in a changing environment
    Biological Reviews of the Cambridge Philosophical Society 96:1854–1867.
    https://doi.org/10.1111/brv.12728
  73. Software
    1. Therneau T
    (2015)
    A package for survival analysis in S
    Version 2.3.8, R Package.

Decision letter

  1. George H Perry
    Senior and Reviewing Editor; Pennsylvania State University, United States
  2. Ivain Martinossi
    Reviewer; Uppsala University, Sweden
  3. Lennart Winkler
    Reviewer; TU Dresden, Germany

Our editorial process produces two outputs: (i) public reviews designed to be posted alongside the preprint for the benefit of readers; (ii) feedback on the manuscript for the authors, including requests for revisions, shown below. We also include an acceptance summary that explains what the editors found interesting or important about the work.

Decision letter after peer review:

Thank you for submitting your article "Thermal phenotypic plasticity of pre-and post-copulatory male harm buffers sexual conflict in wild Drosophila melanogaster" for consideration by eLife. Your article has been reviewed by 3 peer reviewers, and the evaluation has been overseen by a Reviewing Editor and George Perry as the Senior Editor. The following individuals involved in the review of your submission have agreed to reveal their identity: Ivain Martinossi (Reviewer #1); Lennart Winkler (Reviewer #2).

The reviewers have discussed their reviews with one another, and the Reviewing Editor has drafted this to help you prepare a revised submission.

Essential revisions:

1. A consistent critique from all reviewers concerned the need for improved clarity regarding experimental design, both in terms of the experimental steps themselves and also in how the experiments themselves explicitly relate back to the phenomena being studied. Numerous general and specific comments on this point are provided in the below reviews. Please also consider developing one or more figures to assist readers at a high level (while also not neglecting the need for major improvement in clarity and thoroughness at deeper levels in the text itself).

2. The above comments can be extended to the statistical frameworks applied in the paper, i.e. requiring more clarity, depth, and precision (and correction for multiple tests as appropriate in some cases). In some cases, this will likely require adjustment to the statistical approach used for a given hypothesis test.

3. Consider tempering slightly your conclusion that the effect of sexual conflict can be buffered by temperature in the wild, based on the experiments conducted to date.

Reviewer #1 (Recommendations for the authors):

Partly because the methods are placed at the end in this format, I had a lot of confusing moments going through the results. Certain details need to be explained before the methods section. For example, the fact that the polygamy treatment is also a highly male-biased treatment is very important and should be stated clearly. This will impact the interpretation of the results. It is also very important that you explain better the experimental design before diving into the results, otherwise much confusion follows. To be fair, even going through the methods section I found it quite hard to understand the details of the experimental design, especially the receptivity experiment part. It is a complicated experiment with a lot of aspects to it. It needs to be explained very carefully. Perhaps a diagram could help? In any case, you need to explain better the general architecture of the design before getting into the result section. Here are the facts that I think need to be mentioned before the result section so the results can be understood at all (either end of the introduction or the first paragraph of the Results section):

– There are several experiments in parallel, not just one, all across 3 temperatures.

– In experiment 1, flies are exposed to three temperatures in either monogamy (1:1) or polygamy (1f:3m) conditions. LRS is measured, senescence too, and behavior (courtship and aggression) on day 1 are recorded.

– In experiment 2, flies are also exposed to 3 temperatures. This part is very confusing in the methods l.508-525 and after reading it 5 times and drawing diagrams on my whiteboard I am still very unsure of what happened. As far as I got it, virgin individuals are collected and then paired repeatedly in monogamous assays. Males can either come from a "high sperm competition" pool (they are stored together with other virgin males) or "low competition" (males stored alone). There is also a duration factor which I understand to be how long the males sit in their respective competition and temperature treatments before the mating assays. What is clear is that this experiment leads to mating duration and latency measurements (the fact that this is not part of the "behavior" assays is also confusing while going through the results).

– Experiment 3, with the same male treatments as experiment 2, this time female fecundity and egg survival are measured. (where do we see the data from that experiment? As far as I could tell all the figures and supplementary figures are from experiments 1 and 2).

I apologize for the long review. I think it partly reflects my difficulty going through the results and methods section. Overall, I really liked the study even though I was frustrated at times by how much effort I had to put to understand certain figures or results. I think this can become a great manuscript if the methods and results are given just a bit more structure. I would also like to see some slight changes to the statistical methods and a couple of additional supplementary figures (female fecundity data and egg survival). See my details comments below.

Abstract

In the second mention of the mating system treatments, I would reiterate the term polyandry instead of using just "high male competition". Otherwise, it is not immediately clear that it is a mating system treatment. For example "At 20C, female senescence was accelerated under polyandry (high male competition)" and "At 28C, polyandry mostly resulted in reproductive ageing". As far as I understand these results come from Experiment 1 (following my numbering) so they do not correspond to the "male sperm competition" treatments of Experiments 2 and 3.

Not entirely clear results in the abstract. What does it mean when the authors say that at 28C female reproduction is "modulated"?

Results

This is a very dense result section with a lot of interesting data! I would suggest using post hoc tests to investigate the interaction effects, instead of running separate models per temperature. This way, you get p-values corrected for multiple testing. Regarding male-male aggression, I do not think it is correct to assign a zero for the monogamy treatments and test for interaction. The behavior is simply impossible in that treatment, and what you are really looking at is the main effect of temperature on the behavior in polyandry settings. Presenting it as you do is very confusing both in the text and Figure 4 and, I think, is not correct. To clarify the structure further, I would not use exactly the same colors for Polygamy/monogamy and high/low sperm competition. See detailed comments below. I also suggest clearly separating the results that come from different experiments so the readers know which sections are>

l.101: Maybe "temperature by mating system" instead of "x mating system" would be better when inline.

l.103: I would suggest running a post hoc test to see at which temperature the two mating systems differ or not, for example, Tukey's post hoc, instead of three separate models per temperature. What you do is also okay and I have no doubt that this is what the data is showing given how clear it is in Figure 1. But a post hoc would be nicer than 3 separate models because then you get proper p-values corrected for multiple testing.

You could use the package "emmeans" in R. See code below:

>library(emmeans)

>model<-lm(LRS~treatment*temperature, data=data)

>pairs(emmeans(model, ~temperature|treatment, adjust="Tukey"))

l.106: what is "H"?

l.107-109: I am confused about this part and the figure that goes with it. Where do those estimates come from? How are they calculated/simulated? See further comments below in the Figure section.

l.110: not exactly clear what model is run there. Is it only on time point 1 of Figure SI1? Please be more specific.

l.127-135: It is a bit difficult to interpret the model with its simplified estimate of reproductive senescence in the light of figure SI1 which shows the whole complexity of the data. It would be helpful to also have a figure with the mean (week 1, week 2) and mean (week 3, week 4). Do you justify why leaving out the last time point in the methods section?

l.130-131: It may be clearly significant, but it is not clear from the figure which way the effect is going. Perhaps if you provide a simplified figure with means it would help a bit. Also, same as above if you run post hoc tests you will have correct p-values and it should give you an estimate for the differences. The same comment goes for actuarial senescence.

l.147: aggression rate of…I am going to guess males towards females. Please specify. It could also be male-male (or female-female, at least in my study system…), but there can't be male-male aggression in a monogamy assay since there should be only one male. Same as above, please use post hoc tests to determine how the interaction is playing out.

l.147 again: I am super confused now, given that Figure 4a shows only 1 color (no monogamy values) and the figure legend says that we are looking at male-male aggression. Of course, male-male aggression should only be present in the polygamy treatment, but then what is the "estimated decrease"? is it compared to the zero of the monogamy treatment? And what is the interaction term in the model? Did you specify "zero male-male" aggression events in all the monogamy treatment observations? I don't think this is correct to use the monogamy treatment as a comparison here and to fit an interaction effect. What you are really looking at is the main effect of temperature on male-male aggression in polygamy treatments.

L.157: It would be interesting to also look at the rejection rate relative to the courtship rate. For example, you could see that in monogamy at 24C rejection rate seems to increase while the courtship rate decreases…probably not significant but still, another way to look at it that should be informative. Do females reject more, or do they just reject the same in proportion to courtship intensity?

l.162: I find it rather strange to title that section "ejaculate effects" since no ejaculates are sampled, weighed, or analyzed in any way. I understand that the assumption behind this setup is that there is male harm that can happen through ejaculate toxicity, but it is too much of an interpretation to bring that up here. What you measure is female mating behavior and female life history traits (fecundity, egg survival), the title of the section should reflect that.

l.165: Up to that point and before reading the methods section I was not even aware that there was a parallel experiment running with different treatments! That's not good. I got so lost the first time I encountered that part of the result section!! You need to (i) introduce better the experimental design and (ii) structure better the result section so we know where we are. Just the appearance of that "treatment duration" is confusing. If I understood correctly, it is the time that males are exposed to their respective temperature and competition treatments. If that is the case, why is the data not structured that way in Figure 5? We only see the different temperatures and "sperm competition" treatments, the duration of exposure is absent, which does not help.

L170: The same comment for 5b, why does the figure not represent the data, especially given the significant interaction between temperature and treatment duration?

l.179: I would like to see this data represented in a figure (supplementary is fine), including the treatment duration effect.

l.179: Just to emphasize again how it is absolutely needed to clarify the design and structure of the results: any reader who arrives at that point of the text and reads "for the number of eggs produced by females during the first 3 days…." Will assume that we are still talking about the same females as from the LRS data in the beginning and be utterly confused.

l.192: There is an effect of treatment duration on egg/offspring number and survival, but we do not even know in which direction. Please support with figures or give model estimates.

Discussion

I feel that the results from the different experiments remain too separated in the discussion. It is the opportunity to connect together the different pieces, and there are many in this complex puzzle. Is the behavior data consistent with the trend in the LRS data? The lifespan data? The fecundity data? There are some attempts in the last section but I feel that more can be done. The conclusion feels very generic.

l. 202: please remind the reader "net harm in terms of reduced LRS".

L204-207: after reading the methods, there are still no explanations about how those are calculated.

l.210: I would replace "Thus" with "More specifically".

l. 211: Did cold temperature increase female senescence? This is not what I would expect, and not what I see in Figure 3. Females live longer when it's cooler. The effect of the mating system is stronger at 20C though.

l. 212: modulated? In what way? Again, I think it is a bit of a leap to call that "ejaculate effects". Yes, it is likely part of the mechanism, but no you have not measured any ejaculate traits. I would be okay with a statement like "Male competition status affected female mating behavior and fecundity, likely through ejaculate trait. These effects were in turn affected by temperature".

l. 214: if the reproductive senescence is from the separate experiment on fecundity and egg survival, we still need to see the data somehow to be convinced of that statement.

l.236-241: this is actually the best explanation of the rate-sensitive estimates. It comes a bit late and still is not detailed enough.

l.303: ok but what about 28C? the male aggression is at the highest, courtship rate as high as 24C…yet no detectable male harm.

l.311-322: This info is super helpful and should come earlier to introduce this part of the experimental design.

l. 324: with respect to…

l.330-345: Very interesting. Is it possible that females are simply optimizing their mating rates by controlling the remating latency, instead of it depending entirely on the efficiency of SFPs? There are two sides to sexual conflict after all.

l.366: male harm, yes, but female behavior may also be plastic. For example, why see the female rejection rate as a consequence of the action of male SFP, rather than adaptive female behaviors? With this view, sexual selection is only between males, females are just part of the background… that's missing half of the picture.

l.371: I think you are over-simplifying here. Male harm is defined as a reduction in female LRS. At 20C there is a smaller decrease in LRS than at 24C, so reduced male harm, but a larger impact of polyandry on female survival. However, this reduction in survival need not be male harm, since it does not imply a reduction in fitness. It could be part of the female reproductive strategy.

Methods

l.483: I would remove the quotation marks on toxicity.

l.420: polyandry = biased sex ratio as well. Why? Monogamy vs polyandry could be 1:1 vs 3:3. The treatments are effectively "Monogamy" and "Polyandry with a highly male-biased sex ratio". This likely magnifies the effects of sexual conflict. If you were measuring male fitness, there would be additional problems such as underestimating selection on males. I think the sex-ratio unbalance needs to be mentioned more clearly in the first part of the manuscript. It is not an obscure detail to have only in the method section.

Figures

Figure 2: (a) female? Male reproductive success? Just from the result section and the figures, it is very difficult to understand what this is. Does it come from a simulation? Why do the values on the y-axis look so different from figure 1? If it is female LRS, why does it decrease with the population growth rate? So many questions and the few lines in the methods (l.572-577) do not answer them. Consider writing a supplementary page about how this works. (2b) the name of the y-axis is confusing. The cost is obviously highest at 24C but this is the lowest value on that axis so it is the opposite of a cost. Consider changing the variable to 1-X or changing the name. I would also avoid cutting the y-axis if possible (if you change to 1-X and still call it a cost, there is no need to show the zero so you can have your plot centered on the data without cutting the axis).

Figure 4: I understand the need to keep the figures compact but at first glance, the use of two different axes per figure is a bit of a headache. In the end, it would be alright not to have the "estimated decreases" in the main figures, because it is quite visible from the confidence intervals when the two treatments are different. But it is a matter of taste. Also, in Figure 4a it is a bit surprising to still see the estimated decrease when only one treatment (polyandry) is present. What is it compared to?

Figure 5: Keeping the same color schemes for the two experimental designs (polyandry/monogamy and high/low sperm competition) is confusing.

Reviewer #2 (Recommendations for the authors):

Thank you for the opportunity to read this exciting manuscript.

I found the methods detailed, yet I feel that at some points there is still crucial information missing that would be important to judge the robustness of results or for reproducibility (see detailed comments). Furthermore, while I applaud the authors for the great amount of data they collected and the experiments they combine here, I feel that the reader could benefit from a graphic experimental scheme or so to illustrate the procedures. I believe this might help to ease the digestion of the abundant amount of information necessarily given in the methods.

Detailed comments

– L90f: Could you provide more information regarding the source and year these temperature data were collected from?

– L105ff: Were the different harm levels significantly different from each other, as the abstract seems to suggest? I think here only the difference between monogamy and polygamy was tested within temperatures in a pair-wise manner, right?

– L107: 'Rate-sensitive fitness estimates…' Maybe add a bit more context here for an explanation of this term and why this was tested. (It has not been mentioned in the introduction, I think, so this comes a bit out of the blue.)

– Figure 1: I am confused about the model estimate at 20°C. The effect size is smaller compared to 28{degree sign}C and there seems to be just a difference in the CI between those two. This does not fit the data, I believe.

– Figure 2B: I think having a line at 1.0 for no relative fitness cost and then having a break in the y-axis is a bit misleading.

– L147: Male 'aggression rate'?

– L203: '…is not statistically detectable…' That is only a matter of statistical power, isn't it? Unless there is really zero difference, which seems unlikely given the presented data, I believe.

– Figure 4A: What is the comparison plotted here? 'Decrease' from where? In the other figures, it is the comparison between monogamy and polyandry, right?

– L330: '…this effect was consistent for males treated for 48h and 13 days…' Where is this shown? Wasn't there a 'sperm competition risk level x treatment duration interaction' (L164)?

– Figure 5: Are the data for the two experiments pooled here (i.e. 48h and 13 days)? Are these really comparable as the duration of exposure is so different and also the experimental protocol varies (i.e. emptied the seminal fluid and age difference between focal and competitors)?

– L373f: '…maintaining genetic variation in sexually selected traits in males…' I think this is true, but it has not been mentioned before and could do with a bit of context, I believe.

– L406: Could you specify how the temperature fluctuations were achieved? What type of device/incubator was used? Were these fluctuations random or in predictable/pre-programmed way?

– L413: What was the 'controlled density'?

– L415: Please specify, were all collected individuals in separate vials?

– L448: Please clearly define the 'W' in this formula.

– L460f: I am slightly confused by this sentence: 'However, all flies were 5 days old at the start of the experiment.' Do you mean the experiment in general? Then I am not sure what this sentence is here for. Or, do you mean by the start of each behavioral essay? Then I don't understand how the order could have been randomized. Please clarify.

– L522: What exactly does 'right-censored' mean for your analysis?

– L537: How exactly was this modelled? I am not sure that this is sufficient information to be repeatable.

– L550: Maybe add a citation for the R packages used in the analysis?

– Unfortunately, I was not able to access the data or the code via the provided link. Please make sure these are working. I greatly appreciate the publication of code and data.Reviewer #3 (Recommendations for the authors):

I have a number of recommendations that authors should address before the paper is suitable for publication (in the order in which they arise in the manuscript).

1) The authors should avoid detailing the relative harm index H in the abstract. This index will have no meaning to most readers and requires them to read the details of the methods to understand, which is contrary to the purpose of the abstract.

2) L101: Here and throughout the paper it is unclear what statistical models are being tested and how. For example, here the authors test for a significant interaction between temperature and mating system on LRS, and report a chi-square statistic. In the methods, they state that they test the "compared GLMs with their corresponding null GLMs using likelihood ratio test" (L559), so I am guessing that this is the outcome of the test. What is the null model, however? Is it the model without the interaction and just the main effects? They are not explicit here. Further, tests of the main effects are generally considered unimportant when adjusting for the interaction (although trends in the main effects can nevertheless be interpreted). Did they adjust for the interaction when testing for the main effects? Again, this is unclear from the statistical details. I did not have access to the R script (which is stated as being available on Dryad but appears not to be), so am unsure as to what the tests actually are. It may be more straightforward to report the results of the GLMs with an ANOVA table.

3) The authors subsequently test the effect of the mating system on LRS at each temperature and report the harm index. Since these are posthoc tests, they should apply a Bonferoni correction to the significance levels (which may render the effect of the mating system on LRS at 20°C non-significant.

4) Figure 1. Here and in many of the other figures, the authors report both the means (left y-axis) and the contrasts (right y-axis), the latter being labelled 'estimated decrease in LRS'. These estimated decreases look like they are the parameter estimates from the GLMs, but their value depends on how the data are coded. Here they are the decrease in LRS in the polyandrous versus the monogamous, but they could equally be the increase in LRS in the monogamous versus the polyandrous if the authors had coded their data differently. Thus simply labeling the axis 'estimated decrease in LRS' is not sufficient. I think that these parameter values would be best presented in a table, rather than included in the figure. It would be also preferable if the authors included all the data in the figures, for example using a violin plot, rather than just the means and the 95% Cis, since this provides the reader additional information about the distribution of the data.

5) Figure 2. This is perhaps a lack of familiarity on my part, but I do not know how these figures were generated or what they show. These are presumably the outputs of a model (there are some details provided at L575-577) but much more detail needs to be given here.

6) L136. I appreciate that the authors want to provide the details of the statistical analyses on female survival, but all the parentheses make this paragraph extremely difficult to follow. Perhaps including the results of the analysis as a table would make this much easier to read.

7) L146. Again, this paragraph is complex and difficult to follow. The authors may want to state the main finding of their data, before detailing the statistics that support this finding. As for the analysis of the LRS, the P-values of the main effects are questionable if they have been adjusted for significant interaction (which is what is implied in the use of type III ANOVA, L560). As an alternative the authors could conduct posthoc tests on the effect of temperature or mating system for each mating system or temperature respectively, applying appropriate corrections for a multiplicity of tests, e.g. Bonferroni. As for Figure 1, the parameter values for the main effects in Figure 4 (right y-axis) would be best presented in a supplementary table.

8) I find the evidence that temperature affects ejaculate quality to harm females much less convincing. The authors report data on the effects of the mating system on mating duration and remating latency in females mating with males that have been kept singularly (low sperm competition) or in groups of three (high sperm competition). They should clearly explain the relationship between sperm competition, ejaculate quality, and these two assays, with citations if these methods have been used before. The description of these experiments is particularly difficult to follow, but it appears that they recorded mating duration both for the first mating (with males having just been exposed to different levels of perceived sperm competition) and during rematings. Which of these is shown in Figure 5A? They also conducted mating assays on males that had been maintained at different temperatures for 48h or 13 days (L 502). Data from which of these are used in Figure 5A? Which of these were used for female remating latency? Because the experimental methods are so difficult to follow, it precludes an interpretation of the data and makes it difficult to determine whether the data support the conclusions of the authors.

9) L179: This is another dense paragraph that essentially shows that sperm competition risk does not affect female fecundity and survival. That is, while sperm competition may increase mating duration, or remating latency, these do not appear to result in female harm. This undermines the interpretation that changes in mating duration and remating latency with risk of sperm competition reflect the chemistry of the ejaculate. An alternative explanation is that maintaining males with other males may increase his mating duration, which in turn leads to an increase in remating latency. It is interesting to note that the interaction between temperature and risk of sperm competition on mating duration and remating latency reflects a decline in these dependent factors with temperature when risk is low, but maintenance when risk is high. Chemical analysis of the ejaculate would help clarify the relationship between temperature and sperm competition on post-copulatory female harm.

https://doi.org/10.7554/eLife.84759.sa1

Author response

Reviewer #1 (Recommendations for the authors):

Partly because the methods are placed at the end in this format, I had a lot of confusing moments going through the results. Certain details need to be explained before the methods section. For example, the fact that the polygamy treatment is also a highly male-biased treatment is very important and should be stated clearly. This will impact the interpretation of the results. It is also very important that you explain better the experimental design before diving into the results, otherwise much confusion follows. To be fair, even going through the methods section I found it quite hard to understand the details of the experimental design, especially the receptivity experiment part. It is a complicated experiment with a lot of aspects to it. It needs to be explained very carefully. Perhaps a diagram could help? In any case, you need to explain better the general architecture of the design before getting into the result section. Here are the facts that I think need to be mentioned before the result section so the results can be understood at all (either end of the introduction or the first paragraph of the Results section):

Excellent points/suggestions. See responses below.

- There are several experiments in parallel, not just one, all across 3 temperatures.

- In experiment 1, flies are exposed to three temperatures in either monogamy (1:1) or polygamy (1f:3m) conditions. LRS is measured, senescence too, and behavior (courtship and aggression) on day 1 are recorded.

- In experiment 2, flies are also exposed to 3 temperatures. This part is very confusing in the methods l.508-525 and after reading it 5 times and drawing diagrams on my whiteboard I am still very unsure of what happened. As far as I got it, virgin individuals are collected and then paired repeatedly in monogamous assays. Males can either come from a "high sperm competition" pool (they are stored together with other virgin males) or "low competition" (males stored alone). There is also a duration factor which I understand to be how long the males sit in their respective competition and temperature treatments before the mating assays. What is clear is that this experiment leads to mating duration and latency measurements (the fact that this is not part of the "behavior" assays is also confusing while going through the results).

- Experiment 3, with the same male treatments as experiment 2, this time female fecundity and egg survival are measured. (where do we see the data from that experiment? As far as I could tell all the figures and supplementary figures are from experiments 1 and 2).

We agree with the referee that all the facts pointed out need to be clarified before the Results section. To do so, we have reworded and reorganized both the methods and results. More specifically, we have moved the methods to precede the results, and added diagrams for all experiments (Figures S1.1 to S1.5) and a general figure (Figure 1) that reflects the overall design. We now also describe the methods and results independently for each experiment. Additionally, we now provide additional figures in the supplementary material that include the egg-to-adult viability data.

I apologize for the long review. I think it partly reflects my difficulty going through the results and methods section. Overall, I really liked the study even though I was frustrated at times by how much effort I had to put to understand certain figures or results. I think this can become a great manuscript if the methods and results are given just a bit more structure. I would also like to see some slight changes to the statistical methods and a couple of additional supplementary figures (female fecundity data and egg survival). See my details comments below.

Again, thanks for the effort. We agree with all these suggestions. Details follow on how we have dealt with them.

Abstract

In the second mention of the mating system treatments, I would reiterate the term polyandry instead of using just "high male competition". Otherwise, it is not immediately clear that it is a mating system treatment. For example "At 20C, female senescence was accelerated under polyandry (high male competition)" and "At 28C, polyandry mostly resulted in reproductive ageing". As far as I understand these results come from Experiment 1 (following my numbering) so they do not correspond to the "male sperm competition" treatments of Experiments 2 and 3.

Not entirely clear results in the abstract. What does it mean when the authors say that at 28C female reproduction is "modulated"?

We now use the term polyandry to explain our results in the abstract and also re-phrased the sentences to make them clearer. “At 20°C, male harassment of females was reduced, and polyandry accelerated female actuarial ageing. In contrast, the effect of mating on female receptivity (a component of ejaculate toxicity) was affected at 28ºC, where the mating costs for females decreased and polyandry mostly resulted in accelerated reproductive ageing”.

What we meant with “at 28°C female reproduction is modulated” is that, as we now describe in the abstract, female receptivity was only affected by male sperm competition risk at 28°C.

Results

This is a very dense result section with a lot of interesting data! I would suggest using post hoc tests to investigate the interaction effects, instead of running separate models per temperature. This way, you get p-values corrected for multiple testing.

As we now explain in the manuscript (lines 314-324), we did not use Tukey contrasts for our original models because we modelled temperature as a continuous variable (which it is). However, we now use the Benjamini-Hochberg method to control for multiple testing. To err on the side of caution, we also re-run our analyses using temperature as a categorical factor, which allowed us to run Tukey contrasts as an additional way to explore interactions while correcting for multiple testing. Both approaches yield qualitatively identical results and we present the latter models in the SM (tables S1.1, S1.2c, S3.1 and S4.2).

Regarding male-male aggression, I do not think it is correct to assign a zero for the monogamy treatments and test for interaction. The behavior is simply impossible in that treatment, and what you are really looking at is the main effect of temperature on the behavior in polyandry settings. Presenting it as you do is very confusing both in the text and Figure 4 and, I think, is not correct.

The reviewer is right. We re-run the model looking at male-male aggression as suggested and report these results in the current version (lines 415-416). “For male-male aggression rate we detected a significant temperature effect (F1,214 = 14.45; p = 0.0003 Figure 5c and S5.1c)”.

To clarify the structure further, I would not use exactly the same colors for Polygamy/monogamy and high/low sperm competition. See detailed comments below. I also suggest clearly separating the results that come from different experiments so the readers know which sections are:

We agree with this and have used different colours for polyandry/monogamy and high/low sperm competition risk experiments on the respective figures. More specifically, on figures 2 to 5 (and Figures S2.1-2, S4.1 and S5.1-2) we used orange and violet to differentiate between polyandry and monogamy, respectively, and on figure 6 (and Figures S6.1-4) we used yellow and blue to differentiate between high vs low sperm competition risk. We have also separated the results by experiments.

l.101: Maybe "temperature by mating system" instead of "x mating system" would be better when inline.

Done. Throughout all the descriptions of the results we have used “by” instead of “x” to report the interactions between effects.

l.103: I would suggest running a post hoc test to see at which temperature the two mating systems differ or not, for example, Tukey's post hoc, instead of three separate models per temperature. What you do is also okay and I have no doubt that this is what the data is showing given how clear it is in Figure 1. But a post hoc would be nicer than 3 separate models because then you get proper p-values corrected for multiple testing.

You could use the package "emmeans" in R. See code below:

>library(emmeans)

>model<-lm(LRS~treatment*temperature, data=data)

>pairs(emmeans(model, ~temperature|treatment, adjust="Tukey"))

Thank you for the suggestion and the code. As we explained before, we originally did not use Tukey contrasts for our original models because we modelled temperature as a continuous variable. However, we also re-run our analyses using temperature as a categorical factor, which allowed us to run Tukey contrasts. We present those results in the SM (see tables S1.1, S1.2c, S3.1 and S4.2).

l.106: what is "H"?

Clarified in the methods section (lines 177-181) before it appears in the results and discussion. “…by calculating the relative harm (H) following Yun et al. (2021):…”

l.107-109: I am confused about this part and the figure that goes with it. Where do those estimates come from? How are they calculated/simulated? See further comments below in the Figure section.

In the current version (lines 188-201), we detail how we calculate rate-sensitive fitness estimates. Briefly, based on our LRS measures we calculated both individual (ωind) and population (ωpop) ratesensitive estimates for different intrinsic rates of population growth (r) following the procedure detailed in Edward et al. 2011. We also calculated the relative cost of polyandry for each temperature treatment across different values of r, using the population rate-sensitive estimate.

l.110: not exactly clear what model is run there. Is it only on time point 1 of Figure SI1? Please be more specific.

We partitioned overall LRS effects into effects on early reproductive rate (i.e., offspring produced during the first two weeks of age), actuarial ageing (i.e., lifespan), and reproductive ageing (i.e., offspring produced over weeks 1-2 vs. 3-4) that we analysed separately fitting generalized linear models. We also re-made the figure (Figure S2.2), which now shows the mean of each component. This is explained in lines 182-184 and 325-329.

l.127-135: It is a bit difficult to interpret the model with its simplified estimate of reproductive senescence in the light of figure SI1 which shows the whole complexity of the data. It would be helpful to also have a figure with the mean (week 1, week 2) and mean (week 3, week 4). Do you justify why leaving out the last time point in the methods section?

Agree. We have re-made figure SI1 (now Figure S2.2) with the changes suggested by the reviewer. See lines 185-187 in the manuscript for justification about not using week 5 to calculate late reproductive rate and reproductive ageing. “We used weeks 3-4 as an estimate of late reproductive rate because mortality was already evident at this point (i.e., reflecting ageing) and then was very high from week 5 onwards (Figure S4.1; thus preventing accurate estimation of reproductive success)”.

l.130-131: It may be clearly significant, but it is not clear from the figure which way the effect is going. Perhaps if you provide a simplified figure with means it would help a bit. Also, same as above if you run post hoc tests you will have correct p-values and it should give you an estimate for the differences. The same comment goes for actuarial senescence.

As suggested, we have implemented that change (Figure S2.2) and provided the estimates for the differences (tables 1 and S1.1a).

l.147: aggression rate of…I am going to guess males towards females. Please specify. It could also be male-male (or female-female, at least in my study system…), but there can't be male-male aggression in a monogamy assay since there should be only one male. Same as above, please use post hoc tests to determine how the interaction is playing out.

Totally agree. We clarify this in the text and re-run the model looking at male-male aggression only for the polyandry mating system (lines 350-352). We report these results in the current version (lines 415-416).

l.147 again: I am super confused now, given that Figure 4a shows only 1 color (no monogamy values) and the figure legend says that we are looking at male-male aggression. Of course, male-male aggression should only be present in the polygamy treatment, but then what is the "estimated decrease"? is it compared to the zero of the monogamy treatment? And what is the interaction term in the model? Did you specify "zero male-male" aggression events in all the monogamy treatment observations? I don't think this is correct to use the monogamy treatment as a comparison here and to fit an interaction effect. What you are really looking at is the main effect of temperature on male-male aggression in polygamy treatments.

We apologize for the misunderstanding. As explained in the previous comment, we have now run the model checking for male-male aggression only for the polyandry mating system (lines 350-352). Now the figure does not show any estimated decrease (Figure 5c).

L.157: It would be interesting to also look at the rejection rate relative to the courtship rate. For example, you could see that in monogamy at 24C rejection rate seems to increase while the courtship rate decreases…probably not significant but still, another way to look at it that should be informative. Do females reject more, or do they just reject the same in proportion to courtship intensity?

Thank you for the suggestion. We have implemented an analysis to explore potential differences in female rejections per courtship (see results in lines 413-415). “We did not detect a significant interaction between temperature and mating system (F1,294 = 0.30; p = 0.582), nor a temperature effect (F1,294 = 0.08; p = 0.773), for rejection rates per courtship.”

l.162: I find it rather strange to title that section "ejaculate effects" since no ejaculates are sampled, weighed, or analyzed in any way. I understand that the assumption behind this setup is that there is male harm that can happen through ejaculate toxicity, but it is too much of an interpretation to bring that up here. What you measure is female mating behavior and female life history traits (fecundity, egg survival), the title of the section should reflect that.

We have changed “ejaculate effects” by “mating effects”.

l.165: Up to that point and before reading the methods section I was not even aware that there was a parallel experiment running with different treatments! That's not good. I got so lost the first time I encountered that part of the result section!! You need to (i) introduce better the experimental design and (ii) structure better the result section so we know where we are. Just the appearance of that "treatment duration" is confusing. If I understood correctly, it is the time that males are exposed to their respective temperature and competition treatments. If that is the case, why is the data not structured that way in Figure 5? We only see the different temperatures and "sperm competition" treatments, the duration of exposure is absent, which does not help.

For better understanding, as we explained in previous comments, the methods section is now placed before the results. We re-structured both sections, and extensively revised the content to address all the issues that were raised by the reviewer. We also added diagrams for all experiments (Figures S1.1 to S1.5) and an overall figure that outlines the design of the study (Figure 1). We also re-made figure 5 (now is the figure 6) to include both treatment durations. We also supply a table with the estimates for the differences (table 3).

L170: The same comment for 5b, why does the figure not represent the data, especially given the significant interaction between temperature and treatment duration?

Initially, we combined the data from two experiments, which lasted 48 hours and 13 days, solely to demonstrate the general impact of sperm competition risk, as it remained consistent across both treatment durations. However, in the present version of our manuscript, we have created a new figure 6 (formerly figure 5), which shows the results from each experiment separately.

l.179: I would like to see this data represented in a figure (supplementary is fine), including the treatment duration effect.

Done. Please, see figures S6.2 and S6.3.

l.179: Just to emphasize again how it is absolutely needed to clarify the design and structure of the results: any reader who arrives at that point of the text and reads "for the number of eggs produced by females during the first 3 days…." Will assume that we are still talking about the same females as from the LRS data in the beginning and be utterly confused.

Again, thank you for the suggestion. We have reorganized the methods and Results sections and made significant revisions to the content in order to address the aforementioned point. We have separated the methods and results by experiments and directly pointed them out at the beginning of each section.

l.192: There is an effect of treatment duration on egg/offspring number and survival, but we do not even know in which direction. Please support with figures or give model estimates.

Done. Please, see figures S6.2 and S6.3 and tables 4 and S4.1.

Discussion

I feel that the results from the different experiments remain too separated in the discussion. It is the opportunity to connect together the different pieces, and there are many in this complex puzzle. Is the behavior data consistent with the trend in the LRS data? The lifespan data? The fecundity data? There are some attempts in the last section but I feel that more can be done. The conclusion feels very generic.

We have re-written some parts of the discussion to better integrate all the different results. We divided the discussion in two main sections: “Temperature effects on male harm and its consequences for populations” and “Temperature effects on sexual conflict mechanisms in Drosophila melanogaster”. Across the latter, we discuss whether the behaviour (lines 586-596) and fecundity (lines 607-610, 626-631) data is consistent with the trend in the LRS data.

l. 202: please remind the reader "net harm in terms of reduced LRS".

Done. “…net harm to females varies markedly within this range, such that relative harm (i.e., the proportional reduction in average female LRS in polygamy vs. monogamy) was…” (lines 471-473).

L204-207: after reading the methods, there are still no explanations about how those are calculated.

In the current version (lines 188-201), we detail how we calculate rate-sensitive fitness estimates.

l.210: I would replace "Thus" with "More specifically".

Done.

l. 211: Did cold temperature increase female senescence? This is not what I would expect, and not what I see in Figure 3. Females live longer when it's cooler. The effect of the mating system is stronger at 20C though.

At cold temperatures polyandry accelerated female actuarial ageing compared with warm temperatures. We re-phrased the paragraph accordingly (lines 480-482).

l. 212: modulated? In what way? Again, I think it is a bit of a leap to call that "ejaculate effects". Yes, it is likely part of the mechanism, but no you have not measured any ejaculate traits. I would be okay with a statement like "Male competition status affected female mating behavior and fecundity, likely through ejaculate trait. These effects were in turn affected by temperature".

We hope this is clear in the current version: “Warm temperatures impacted mating costs and effects on female receptivity (i.e., post-copulatory mechanism), and female fitness decreased more via accelerated reproductive ageing than at colder temperatures” (lines 482-484).

l. 214: if the reproductive senescence is from the separate experiment on fecundity and egg survival, we still need to see the data somehow to be convinced of that statement.

We now present the reproductive senescence data in figure S2.2.

l.236-241: this is actually the best explanation of the rate-sensitive estimates. It comes a bit late and still is not detailed enough.

As suggested, we have gone through the explanation in a more detailed way (lines 188-201).

l.303: ok but what about 28C? the male aggression is at the highest, courtship rate as high as 24C…yet no detectable male harm.

We clarified in lines 593-596 that the decrease in male harm that we detected at 20° vs 24 ° is likely explained by the substantial drop in male harassment at this temperature. However, the same logic cannot apply for the decrease in male harm to females that we detected at 28°C vs. 24°C. Instead, receipt of a male ejaculate resulted in a sharper decrease of female receptivity in high vs. low SCR at 28°C (lines 627-631).

l.311-322: This info is super helpful and should come earlier to introduce this part of the experimental design.

Now it is part of the methodology, and thus comes before explaining the experimental design (lines 239-247).

l. 324: with respect to…

Changed.

l.330-345: Very interesting. Is it possible that females are simply optimizing their mating rates by controlling the remating latency, instead of it depending entirely on the efficiency of SFPs? There are two sides to sexual conflict after all.

Agree. We have extended this idea (lines 636-647).

l.366: male harm, yes, but female behavior may also be plastic. For example, why see the female rejection rate as a consequence of the action of male SFP, rather than adaptive female behaviors? With this view, sexual selection is only between males, females are just part of the background… that's missing half of the picture.

We agree with the fact that temperature may affect female behaviour or physiology in a way that makes them more resistant to harm. We have included the female perspective in the current discussion (lines 636-653).

l.371: I think you are over-simplifying here. Male harm is defined as a reduction in female LRS. At 20C there is a smaller decrease in LRS than at 24C, so reduced male harm, but a larger impact of polyandry on female survival. However, this reduction in survival need not be male harm, since it does not imply a reduction in fitness. It could be part of the female reproductive strategy.

We have discussed the point that the reviewer highlights throughout the last part of the Discussion section (lines 636-647).

Methods

l.483: I would remove the quotation marks on toxicity.

Done.

l.420: polyandry = biased sex ratio as well. Why? Monogamy vs polyandry could be 1:1 vs 3:3. The treatments are effectively "Monogamy" and "Polyandry with a highly male-biased sex ratio". This likely magnifies the effects of sexual conflict. If you were measuring male fitness, there would be additional problems such as underestimating selection on males. I think the sex-ratio unbalance needs to be mentioned more clearly in the first part of the manuscript. It is not an obscure detail to have only in the method section.

We clarify now in the manuscript that the sex ratio in this species is typically 1:1, however the operational sex ratio is male-biased and frequently reaches a 3:1 (or higher) male-bias in mating patches in the wild. Thus, the 1:1 vs. 3:1 sex ratios used in this study represent biologically relevant scenarios and have actually become standard in Drosophila studies measuring male harm to females. See lines 150-155.

Figures

Figure 2: (a) female? Male reproductive success? Just from the result section and the figures, it is very difficult to understand what this is. Does it come from a simulation? Why do the values on the y-axis look so different from figure 1? If it is female LRS, why does it decrease with the population growth rate? So many questions and the few lines in the methods (l.572-577) do not answer them. Consider writing a supplementary page about how this works. (2b) the name of the y-axis is confusing. The cost is obviously highest at 24C but this is the lowest value on that axis so it is the opposite of a cost. Consider changing the variable to 1-X or changing the name. I would also avoid cutting the y-axis if possible (if you change to 1-X and still call it a cost, there is no need to show the zero so you can have your plot centered on the data without cutting the axis).

In the current version, we detail how we calculate rate-sensitive fitness estimates (lines 188-201). We also implemented changes in figure 2 (now figure 3). Panel 3a shows the average fitness estimate of individual females (Wind) for different population growth rates. We have implemented the change suggested by the reviewer for the panel 3b.

Figure 4: I understand the need to keep the figures compact but at first glance, the use of two different axes per figure is a bit of a headache. In the end, it would be alright not to have the "estimated decreases" in the main figures, because it is quite visible from the confidence intervals when the two treatments are different. But it is a matter of taste. Also, in Figure 4a it is a bit surprising to still see the estimated decrease when only one treatment (polyandry) is present. What is it compared to?

We have left the figures only with the means and now provide tables with the estimates to explore size effects.

Figure 5: Keeping the same color schemes for the two experimental designs (polyandry/monogamy and high/low sperm competition) is confusing.

We have changed the colour for the experiments 2 to 5 (high/low sperm competition risk). Please, see this change in figures 6 and S6.1-4.

Reviewer #2 (Recommendations for the authors):

Thank you for the opportunity to read this exciting manuscript.

I found the methods detailed, yet I feel that at some points there is still crucial information missing that would be important to judge the robustness of results or for reproducibility (see detailed comments). Furthermore, while I applaud the authors for the great amount of data they collected and the experiments they combine here, I feel that the reader could benefit from a graphic experimental scheme or so to illustrate the procedures. I believe this might help to ease the digestion of the abundant amount of information necessarily given in the methods.

We completely agree. Please notice that to clarify the methods we have placed this section before the Results section and re-written both the methods and results. This has significantly improved the clarity of the manuscript. We have also included several diagrams to illustrate all our experiments (in the SM. Figures S1.1-5) along with a general schematic figure of the whole design that we present early on (in the introduction, see Figure 1).

Detailed comments

– L90f: Could you provide more information regarding the source and year these temperature data were collected from?

Sure. The data come from WorldClim2, which has average monthly climate data for minimum, mean, and maximum temperature and for precipitation for 1970-2000 (Fick and Hijmans, 2017). Now, it is referenced in line 92.

– L105ff: Were the different harm levels significantly different from each other, as the abstract seems to suggest? I think here only the difference between monogamy and polygamy was tested within temperatures in a pair-wise manner, right?

As we detected a significant interaction between temperature by mating system for LRS, we interpreted that net male harm varies markedly among temperatures. To explore the nature of the interaction we ran models separately for each temperature in which only the difference between monogamy and polyandry is tested. We found a larger effect of mating system on LRS at 24°C than at 20°C, and no effect at 28°C (tables 1 and S1.1a).

– L107: 'Rate-sensitive fitness estimates…' Maybe add a bit more context here for an explanation of this term and why this was tested. (It has not been mentioned in the introduction, I think, so this comes a bit out of the blue.)

In the current version, we detail how we calculated rate-sensitive fitness estimates (lines 188-201).

– Figure 1: I am confused about the model estimate at 20°C. The effect size is smaller compared to 28{degree sign}C and there seems to be just a difference in the CI between those two. This does not fit the data, I believe.

We have double-checked the data and re-run the analysis to be sure about the output of the models. Separate models for each temperature level show an estimate at 20°C = 1.07, 95% CI 0.06 -2.06 and at 28°C = 1.88, 95% CI -0.58 – 4.36 (table 1). However, note that the Tukey’s post hoc test yielded the same results for mating system effect as running the models separately for each temperature, albeit with a greater estimate at 20° than at 28° (table S1.1a-LRS-).

– Figure 2B: I think having a line at 1.0 for no relative fitness cost and then having a break in the y-axis is a bit misleading.

We agree with this. We have re-made figure 2b (now figure 3b) and now the population costs are shown as 1 – Cr for better understanding.

– L147: Male 'aggression rate'?

We meant male aggression rate. Clarified in line 415.

– L203: '…is not statistically detectable…' That is only a matter of statistical power, isn't it? Unless there is really zero difference, which seems unlikely given the presented data, I believe.

We have re-formulated the phrase to avoid misunderstandings (lines 471-473). “…that relative harm (i.e., the proportional reduction in average female LRS in polygamy vs. monogamy) was maximal at 24ºC, decreased at 20ºC and was minimum at 28°C…”.

– Figure 4A: What is the comparison plotted here? 'Decrease' from where? In the other figures, it is the comparison between monogamy and polyandry, right?

We have re-run the model looking at male-male aggression only for the polyandry mating system. We report these results in the current version (lines 415-416). We re-made figure 4 (now figure 5) and you can see that now there is not a comparison plotted for male-male aggression rate (Figure 5c), but for other figures the comparison is between monogamy and polyandry (Figure 5a-b).

– L330: '…this effect was consistent for males treated for 48h and 13 days…' Where is this shown? Wasn't there a 'sperm competition risk level x treatment duration interaction' (L164)?

We re-made figure 5 (now figure 6), and now show both treatment durations in the figure. We also supply a table with the estimates for the differences (Table 3).

– Figure 5: Are the data for the two experiments pooled here (i.e. 48h and 13 days)? Are these really comparable as the duration of exposure is so different and also the experimental protocol varies (i.e. emptied the seminal fluid and age difference between focal and competitors)?

Yes, the data for the two experiments (48h and 13 days) were pooled initially because we wanted to focus on the overall effect of the sperm competition risk, as it was consistent across treatment durations. However, in the current version of the manuscript, we re-made figure 5 (now figure 6) and show both experiments independently.

– L373f: '…maintaining genetic variation in sexually selected traits in males…' I think this is true, but it has not been mentioned before and could do with a bit of context, I believe.

Now this idea is also considered at the first part of the Discussion section (lines 537-546). As a potential consequence arising from the observed temperature effects on male harm mechanisms, we propose that environmental variability may foster the maintenance of genetic variation underlying different mechanisms of male harm, and even potentially divergent male-male competition strategies.

– L406: Could you specify how the temperature fluctuations were achieved? What type of device/incubator was used? Were these fluctuations random or in predictable/pre-programmed way?

We pre-programmed daily temperature fluctuations (24 ±4°C) mimicking natural daily temperature conditions during the reproductively active season using a Pol Eko ST 1200 incubator. We now clarify this in lines 124-126.

– L413: What was the 'controlled density'?

Every time we collected eggs to obtain experimental flies, we used a controlled density of ~200eggs (223 ± 14.3, mean and 95% CI), following the procedure described by Clancy and Kennington, 2001. We now specify this in line 137.

– L415: Please specify, were all collected individuals in separate vials?

Once we collected and sexed virgin experimental flies, the way that we kept them until their use depended on the experiment. For example, for the experiment 1 we collected virgin flies into same-sex vials of 15 individuals (line 156). However, for the experiments 2 to 5 we placed the experimental males in vials either individually (low sperm competition risk) or in a same-sex group of 8 (high sperm competition risk; clarified in lines 264-266) before we allocated them to the different temperature treatments.

– L448: Please clearly define the 'W' in this formula.

To calculate relative harm (H), we followed the formula described by Yun et al. (2021) where W corresponds to female’s fitness. We now clarify this in lines 177-181.

– L460f: I am slightly confused by this sentence: 'However, all flies were 5 days old at the start of the experiment.' Do you mean the experiment in general? Then I am not sure what this sentence is here for. Or, do you mean by the start of each behavioral essay? Then I don't understand how the order could have been randomized. Please clarify.

Indeed, we have worked in the current version to make the method section easier to follow and more comprehensive (lines 202-210). What we meant is that, for experiment 1, we conducted behavioural observations in the same temperature control room, due to logistic limitations. So, we had to conduct trials at 20, 24 and 28°C over three consecutive days (because the same CT room can only be kept at a given temperature at the same time). However, as we collected virgin flies over three consecutive days, we ensured all flies were 5 days-old at the start of the experiment (i.e., when the observations were done at each temperature). We conducted observations first at 20°, then at 28° and finally at 24°C (we haphazardly chose this order).

– L522: What exactly does 'right-censored' mean for your analysis?

It means individuals that are taken into account for demographic analysis until the day they disappear (Kleinbaum and Klein, 2012). We explain these terms in lines 345-346.

– L537: How exactly was this modelled? I am not sure that this is sufficient information to be repeatable.

We have clarified how the fecundity and survival assays were done (lines 293-310).

– L550: Maybe add a citation for the R packages used in the analysis?

Done (lines 343-344).

– Unfortunately, I was not able to access the data or the code via the provided link. Please make sure these are working. I greatly appreciate the publication of code and data.

We apologize for this. We provide the code and data directly just in case the link from Dryad doesn’t work yet.

Reviewer #3 (Recommendations for the authors):

I have a number of recommendations that authors should address before the paper is suitable for publication (in the order in which they arise in the manuscript).

1) The authors should avoid detailing the relative harm index H in the abstract. This index will have no meaning to most readers and requires them to read the details of the methods to understand, which is contrary to the purpose of the abstract.

Agree, we now express this in %.

2) L101: Here and throughout the paper it is unclear what statistical models are being tested and how. For example, here the authors test for a significant interaction between temperature and mating system on LRS, and report a chi-square statistic. In the methods, they state that they test the "compared GLMs with their corresponding null GLMs using likelihood ratio test" (L559), so I am guessing that this is the outcome of the test. What is the null model, however? Is it the model without the interaction and just the main effects? They are not explicit here. Further, tests of the main effects are generally considered unimportant when adjusting for the interaction (although trends in the main effects can nevertheless be interpreted). Did they adjust for the interaction when testing for the main effects? Again, this is unclear from the statistical details. I did not have access to the R script (which is stated as being available on Dryad but appears not to be), so am unsure as to what the tests actually are. It may be more straightforward to report the results of the GLMs with an ANOVA table.

We have amended the statistical analyses part of the manuscript. Briefly, we have used ANOVA type III (via “F” statistic) to compute p-values corrected by the Benjamini-Hochberg method. We also compared GLMs with their corresponding null GLMs in order to test the significance of the independent variables in the full model (omnibus test). As we used ANOVA type III, it is true that the test of the main effects could be considered unimportant. However, we feel interpretation of interactions and main effects is straightforward given the raw data shown in the figures. In addition, we re-run all analysis and conducted Tukey’s post hoc tests on the effect of mating system for each temperature (as well as on the effect of treatment duration for each mating system; see Tables S1-3), when significant interactions were detected. As you will see, results are very consistent with our initial interpretation of interaction and main effects. We are sorry the reviewer could not get access to the R script. We have contacted Dryad but also provide the code and data in the submission, in case the Dryad link hasn’t been amended by the time the reviewer sees this.

3) The authors subsequently test the effect of the mating system on LRS at each temperature and report the harm index. Since these are posthoc tests, they should apply a Bonferoni correction to the significance levels (which may render the effect of the mating system on LRS at 20°C non-significant.

We thank the reviewer for this suggestion. As we now explain in the manuscript (lines 328-338) we repeated all analyses and applied the Benjamini-Hochberg method to control for multiple testing (the best option given we fit temperature as a continuous variable). Furthermore, we also repeated our analyses fitting temperature as a categorical factor and using Tukey contrasts to further examine interactions while accounting for multiple testing. Both methods produced very similar results, and we included the latter approach in the supplementary material (tables S1.1, S1.2c, S3.1 and S4.2).

4) Figure 1. Here and in many of the other figures, the authors report both the means (left y-axis) and the contrasts (right y-axis), the latter being labelled 'estimated decrease in LRS'. These estimated decreases look like they are the parameter estimates from the GLMs, but their value depends on how the data are coded. Here they are the decrease in LRS in the polyandrous versus the monogamous, but they could equally be the increase in LRS in the monogamous versus the polyandrous if the authors had coded their data differently. Thus simply labeling the axis 'estimated decrease in LRS' is not sufficient. I think that these parameter values would be best presented in a table, rather than included in the figure. It would be also preferable if the authors included all the data in the figures, for example using a violin plot, rather than just the means and the 95% Cis, since this provides the reader additional information about the distribution of the data.

Attending the suggestions by two of the reviewers, we have left the figures only with the means and now we provide tables with the estimates. We preferred to leave the main figures showing the mean +/- SEM, considering that in this way differences are easier to interpret. However, we supply violin plots in the SM (Figures S2.1-2, S4.1, S5.1-2 and S6.1-4).

5) Figure 2. This is perhaps a lack of familiarity on my part, but I do not know how these figures were generated or what they show. These are presumably the outputs of a model (there are some details provided at L575-577) but much more detail needs to be given here.

We agree with the lack of clarity about how figure 2 (now figure 3) was generated. In the current version, you can find an extended explanation addressing the aforementioned changes (lines 188201).

6) L136. I appreciate that the authors want to provide the details of the statistical analyses on female survival, but all the parentheses make this paragraph extremely difficult to follow. Perhaps including the results of the analysis as a table would make this much easier to read.

We have included a table to show the output from separate GLMs for each temperature level to explore significant interactions between temperature and mating system effects for lifespan (table 1).

7) L146. Again, this paragraph is complex and difficult to follow. The authors may want to state the main finding of their data, before detailing the statistics that support this finding. As for the analysis of the LRS, the P-values of the main effects are questionable if they have been adjusted for significant interaction (which is what is implied in the use of type III ANOVA, L560). As an alternative the authors could conduct posthoc tests on the effect of temperature or mating system for each mating system or temperature respectively, applying appropriate corrections for a multiplicity of tests, e.g. Bonferroni. As for Figure 1, the parameter values for the main effects in Figure 4 (right y-axis) would be best presented in a supplementary table.

For a better understanding, we have re-structured and reworded the Results section. While we utilized ANOVA type III to report the p-values. This approach cautions against direct interpretation of main effects, but we feel in this case interpretation is straightforward given the data trends depicted in the figures. However, we conducted further analyses by running Tukey's post hoc tests on the impact of the mating system at each temperature, as shown in table S1.1b for reproductive behaviour; and results are in accordance with our interpretation. We now present all the figures without the right y-axis and provide tables to show the size effects (tables 1-3)

8) I find the evidence that temperature affects ejaculate quality to harm females much less convincing. The authors report data on the effects of the mating system on mating duration and remating latency in females mating with males that have been kept singularly (low sperm competition) or in groups of three (high sperm competition). They should clearly explain the relationship between sperm competition, ejaculate quality, and these two assays, with citations if these methods have been used before. The description of these experiments is particularly difficult to follow, but it appears that they recorded mating duration both for the first mating (with males having just been exposed to different levels of perceived sperm competition) and during rematings. Which of these is shown in Figure 5A? They also conducted mating assays on males that had been maintained at different temperatures for 48h or 13 days (L 502). Data from which of these are used in Figure 5A? Which of these were used for female remating latency? Because the experimental methods are so difficult to follow, it precludes an interpretation of the data and makes it difficult to determine whether the data support the conclusions of the authors.

We are sorry that our initial structure was not clear enough to show the overall experimental design. We now provide more detailed explanations regarding the methods and results. We also re-made figure 5 (now figure 6), showing the data for mating duration for the first mating (Figure 6a) and female remating latency (Figure 6b) for the two treatment durations separately.

9) L179: This is another dense paragraph that essentially shows that sperm competition risk does not affect female fecundity and survival. That is, while sperm competition may increase mating duration, or remating latency, these do not appear to result in female harm. This undermines the interpretation that changes in mating duration and remating latency with risk of sperm competition reflect the chemistry of the ejaculate. An alternative explanation is that maintaining males with other males may increase his mating duration, which in turn leads to an increase in remating latency. It is interesting to note that the interaction between temperature and risk of sperm competition on mating duration and remating latency reflects a decline in these dependent factors with temperature when risk is low, but maintenance when risk is high. Chemical analysis of the ejaculate would help clarify the relationship between temperature and sperm competition on post-copulatory female harm.

We have attempted to streamline and clarify the female fecundity and survival part of the manuscript so hopefully the message we wanted to get across is clearer (lines 444-466). We have also clarified our results in relation to the ejaculate effects on female receptivity (lines 423-437). Briefly, sperm competition risk (i.e., keeping males in a group of males vs. isolation) does affect both mating duration and the effects of mating on female re-mating latency, but the key is that these effects appear to be higher at 28ºC than at other temperatures, which suggests modulation of both these variables at warmer temperatures, as pointed out by the reviewer. Clearly, the reviewer is correct in pointing out that these results are intriguing but cannot be interpreted conclusively without analysis of temperature effects on the composition of ejaculates. We discuss this along with potential explanations in the current version, where we have attempted to clarify what we did, why we did it and what this may mean, but erring on the side of caution (lines 633-655).

https://doi.org/10.7554/eLife.84759.sa2

Article and author information

Author details

  1. Claudia Londoño-Nieto

    Ethology Lab, Cavanilles Institute of Biodiversity and Evolutionary Biology, University of Valencia, Valencia, Spain
    Contribution
    Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing – original draft
    For correspondence
    claudia.londonon@gmail.com
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-7408-7327

  2. Roberto García-Roa

    1. Ethology Lab, Cavanilles Institute of Biodiversity and Evolutionary Biology, University of Valencia, Valencia, Spain
    2. Department of Biology, Lund University, Lund, Sweden
    Contribution
    Supervision, Investigation, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-9568-9191

  3. Clara Garcia-Co

    Ethology Lab, Cavanilles Institute of Biodiversity and Evolutionary Biology, University of Valencia, Valencia, Spain
    Contribution
    Methodology
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-4126-5940

  4. Paula González

    Ethology Lab, Cavanilles Institute of Biodiversity and Evolutionary Biology, University of Valencia, Valencia, Spain
    Contribution
    Methodology
    Competing interests
    No competing interests declared

  5. Pau Carazo

    Ethology Lab, Cavanilles Institute of Biodiversity and Evolutionary Biology, University of Valencia, Valencia, Spain
    Contribution
    Conceptualization, Resources, Supervision, Funding acquisition, Validation, Investigation, Visualization, Methodology, Writing - review and editing
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0002-1525-6522

Funding

Ministerio de Ciencia e Innovación (PID2020-118027GB-I00)

  • Pau Carazo

Generalitat Valenciana (AICO/2021/113)

  • Pau Carazo

Ministerio de Educación y Formación Profesional (FJC2018-037058-I)

  • Roberto García-Roa

Marie Sklodowska-Curie Actions (HORIZON-MSCA-2021-PF-01 101061275)

  • Roberto García-Roa

Ministerio de Asuntos Económicos y Transformación Digital, Gobierno de España (PRE2018-084009)

  • Claudia Londoño-Nieto

European Social Fund (ESF Investing in your future)

  • Claudia Londoño-Nieto

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Alejandro Hita for assistance with the experimental procedure. PC was supported by a research grant (PID2020-118027GB-I00) funded by MCIN/AEI/ 10.13039/501100011033 and a research grant AICO/2021/113 from Generalitat Valenciana. RGR was supported by a postdoctoral grant (FJC2018-037058-I) funded by MCIN/AEI/ 10.13039/501100011033 and by a Marie Sklodowska Curie Fellowship (HORIZON-MSCA-2021-PF-01 101061275). CLN was supported by a predoctoral grant (PRE2018-084009) by MCIN/AEI/ 10.13039/501100011033 and by "ESF Investing in your future".

Senior and Reviewing Editor

  1. George H Perry, Pennsylvania State University, United States

Reviewers

  1. Ivain Martinossi, Uppsala University, Sweden
  2. Lennart Winkler, TU Dresden, Germany

Publication history

  1. Received: November 7, 2022
  2. Preprint posted: November 12, 2022 (view preprint)
  3. Accepted: April 26, 2023
  4. Accepted Manuscript published: April 27, 2023 (version 1)
  5. Version of Record published: May 17, 2023 (version 2)

Copyright

© 2023, Londoño-Nieto et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 248
    Page views
  • 55
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Open citations (links to open the citations from this article in various online reference manager services)

Cite this article (links to download the citations from this article in formats compatible with various reference manager tools)

  1. Claudia Londoño-Nieto
  2. Roberto García-Roa
  3. Clara Garcia-Co
  4. Paula González
  5. Pau Carazo
(2023)
Thermal phenotypic plasticity of pre- and post-copulatory male harm buffers sexual conflict in wild Drosophila melanogaster
eLife 12:e84759.
https://doi.org/10.7554/eLife.84759

Categories and tags

Research organism

Further reading

    1. Ecology
    2. Evolutionary Biology

    Sensory conflict disrupts circadian rhythms in the sea anemone Nematostella vectensis

    Cory A Berger, Ann M Tarrant
    Research Article Updated

    Circadian clocks infer time of day by integrating information from cyclic environmental factors called zeitgebers, including light and temperature. Single zeitgebers entrain circadian rhythms, but few studies have addressed how multiple, simultaneous zeitgeber cycles interact to affect clock behavior. Misalignment between zeitgebers (‘sensory conflict’) can disrupt circadian rhythms, or alternatively clocks may privilege information from one zeitgeber over another. Here, we show that temperature cycles modulate circadian locomotor rhythms in Nematostella vectensis, a model system for cnidarian circadian biology. We conduct behavioral experiments across a comprehensive range of light and temperature cycles and find that Nematostella’s circadian behavior is disrupted by chronic misalignment between light and temperature, which involves disruption of the endogenous clock itself rather than a simple masking effect. Sensory conflict also disrupts the rhythmic transcriptome, with numerous genes losing rhythmic expression. However, many metabolic genes remained rhythmic and in-phase with temperature, and other genes even gained rhythmicity, implying that some rhythmic metabolic processes persist even when behavior is disrupted. Our results show that a cnidarian clock relies on information from light and temperature, rather than prioritizing one signal over the other. Although we identify limits to the clock’s ability to integrate conflicting sensory information, there is also a surprising robustness of behavioral and transcriptional rhythmicity.

    1. Ecology
    2. Evolutionary Biology

    A toxin-mediated policing system in Bacillus optimizes division of labor via penalizing cheater-like nonproducers

    Rong Huang, Jiahui Shao ... Ruifu Zhang
    Research Article Updated

    Division of labor, where subpopulations perform complementary tasks simultaneously within an assembly, characterizes major evolutionary transitions of cooperation in certain cases. Currently, the mechanism and significance of mediating the interaction between different cell types during the division of labor, remain largely unknown. Here, we investigated the molecular mechanism and ecological function of a policing system for optimizing the division of labor in Bacillus velezensis SQR9. During biofilm formation, cells differentiated into the extracellular matrix (ECM)-producers and cheater-like nonproducers. ECM-producers were also active in the biosynthesis of genomic island-governed toxic bacillunoic acids (BAs) and self-resistance; while the nonproducers were sensitive to this antibiotic and could be partially eliminated. Spo0A was identified to be the co-regulator for triggering both ECM production and BAs synthesis/immunity. Besides its well-known regulation of ECM secretion, Spo0A activates acetyl-CoA carboxylase to produce malonyl-CoA, which is essential for BAs biosynthesis, thereby stimulating BAs production and self-immunity. Finally, the policing system not only excluded ECM-nonproducing cheater-like individuals but also improved the production of other public goods such as protease and siderophore, consequently, enhancing the population stability and ecological fitness under stress conditions and in the rhizosphere. This study provides insights into our understanding of the maintenance and evolution of microbial cooperation.

    1. Ecology

    Larger but younger fish when growth outpaces mortality in heated ecosystem

    Max Lindmark, Malin Karlsson, Anna Gårdmark
    Research Article

    Ectotherms are predicted to ‘shrink’ with global warming, in line with general growth models and the temperature-size rule (TSR), both predicting smaller adult sizes with warming. However, they also predict faster juvenile growth rates and thus larger size-at-age of young organisms. Hence, the result of warming on the size-structure of a population depends on the interplay between how mortality rate, juvenile- and adult growth rates are affected by warming. Here, we use two-decade long time series of biological samples from a unique enclosed bay heated by cooling water from a nearby nuclear power plant to become 5–10 °C warmer than its reference area. We used growth-increment biochronologies (12,658 reconstructed length-at-age estimates from 2426 individuals) to quantify how >20 years of warming has affected body growth, size-at-age, and catch to quantify mortality rates and population size- and age structure of Eurasian perch (Perca fluviatilis). In the heated area, growth rates were faster for all sizes, and hence size-at-age was larger for all ages, compared to the reference area. While mortality rates were also higher (lowering mean age by 0.4 years), the faster growth rates lead to a 2 cm larger mean size in the heated area. Differences in the size-spectrum exponent (describing how the abundance declines with size) were less clear statistically. Our analyses reveal that mortality, in addition to plastic growth and size-responses, is a key factor determining the size structure of populations exposed to warming. Understanding the mechanisms by which warming affects the size- and the age structure of populations is critical for predicting the impacts of climate change on ecological functions, interactions, and dynamics.