Fly stock establishment and maintenance
Two D. pachea stocks 15090–1698.01 and 15090–1698.02 were retrieved from the San Diego Drosophila Species Stock Center (now The National Drosophila Species Stock Center, College of Agriculture and Life Science, Cornell University, USA) (Additional file 1: Table S1). Flies were maintained in 25 × 95 mm plastic vials containing 10 mL of standard Drosophila medium (60 g/L brewer’s yeast, 66.6 g/L cornmeal, 8.6 g/L agar, 5 g/L methyl-4-hydroxybenzoate and 2.5% v/v ethanol) and a ~ 10 × 50 mm piece of bench protection sheet (Bench guard). As D. pachea requires 7-dehydrocholesterol for proper development [58,59,60], we mixed the medium of each vial with 40 µL of 5 mg/mL 7-dehydrocholesterol, dissolved in ethanol (standard D. pachea food). Flies were kept at 25 °C inside incubators (Velp) with a self-made light installation for a 12 h light:12 h dark photo-periodic cycle combined with a 30-min linear illumination change between light (1080 lm) and dark (0 lm). We used males of stock 15090–1698.02 to generate a new stock with increased proportions of males with short gential lobes. For this, we chose three males with apparently symmetric (aberrant) genital lobes and crossed them with 3–4 sibling virgin females. We repeated the selection with the progeny for a total of 36 generations. Then, we discarded males with clearly visible asymmetric (wild-type) lobes from the progeny for another 14 generations to derive the final stock (selection stock).
Virgin fly selection
Virgin flies at 0–1 day after emerging from the pupa were CO2 anaesthetised on a CO2-pad (INJECT + MATIC Sleeper) under a stereo-microscope Stemi 2000 (Zeiss), separated according to sex and maintained in groups of 20–30 individuals. Males and females were raised until reaching sexual maturity, about 14 days for males and 4 days for females at 25 °C . This allowed us to use virgin individuals in each mating experiment. Males were anaesthetised on the CO2 pad (see above), sorted according to lobe morphology (asymmetric and symmetric lobes) and isolated into single vials at least 2 days before each mating experiment took place.
We used a single virgin female of wild-type stock 15090–1698.01 or 15090–1698.02 and two sibling males of the selection stock that visually differed in lobe length when inspected with a binocular microscope (Additional file 1: Figure S5). This selection was done to increase the average pairwise difference of lobe lengths between the two males. Specimens were introduced into a white, cylindrical mating cell (Additional file 1: Figure S1) with a diameter of 20 mm, a depth of 4 mm and a transparent 1 mm Plexiglas top-cover. Optionally, mating cells were concave with a diameter of 20 mm and a depth of 4 mm at the center (Additional file 3: trials 1–16). Flies were transferred without CO2 anaesthesia using a fly aspirator: a 7 mm diameter silicone tube closed at the tip with cotton and a 1000 µL wide bore (3 mm) micro-tip. Movie recordings were started as soon as the chamber was immediately put under the camera (see below).
All mating trials were carried out inside a temperature and humidity controlled climate chamber at 25 °C ± 0.1 °C and 80% ± 5% or 60% ± 5% (trials 1–15) humidity. We used digital cameras MIRAZOOM MZ902 (OWL) or DigiMicro Profi (DNT) to record copulation and courtship behaviour. The MIRAZOOM MZ902 (OWL) camera was mounted on a modified microscope stand (191,348, Conrad) equipped with a 8-cm LED white light illumination ring (EB-AE-COB-Cover, YM E-Bright) and a platform to hold four individual cylindrical mating cells (Additional file 1: Figure S1). Trial 16 (Additional file 3) was filmed with the OWL camera and a concave shaped mating cell that was put on a flat plastic cover on top of the microscope stand. For trials 1–15, we used concave shaped mating cells that were put onto the stand of the DigiMicro Profi (DNT) camera. Data was acquired with the programs Cheese (version 3.18.1) (https://wiki.gnome.org/Apps/Cheese) or GUVCVIEW (version 0.9.9) GTK UVC (trials 1–15) on an ubuntu linux operating system in webm or mkv format. Up to four trials were recorded in a single movie, which was split after recording and converted into mp4 format with Avidemux 2.6 (http://www.avidemux.org/) to obtain a single movie per experiment.
We waited for a copulation to take place between one of the males and the female (see below). The mating cell was then shortly recovered from the climate chamber in order to remove the non-copulating male with an aspirator and to transfer it into a 2-mL reaction tube filled with 70% ethanol. This was performed after the copulating male appeared to have settled into a stable mating posture on the female abdomen, which was controlled by live-imaging. The copulating male and the female were isolated from the mating cell after copulation had ended and were also isolated into single 2-mL reaction tubes filled with 70% ethanol. Optionally, the female was kept alive for 12–24 h in a vial containing 5–10 mL grape juice agar (24 g/L agar, 26 g/L sucrose, 120 mg/L Tegosept, 20% v/v grape juice, and 1.5% v/v ethanol). The presence of eggs on the plate was systematically checked but never observed. Females were finally sacrificed to prepare spermathecae.
Left lobe surgery
Epandrial lobe surgery was done on 5–6 day-old D. pachea adult males of the selection stock following Rhebergen et al. . Males were anaesthetised on a CO2 pad (see above) and then further immobilized with a small copper wire, which was slightly pressed onto the male abdomen. The left epandrial lobe was shortened to various lengths with micro dissection scissors (SuperFine Vannas Scissors, World Precision Instruments). Flies were let to recover for at least 7 days on standard D. pachea food at 25 °C in groups. No mortality was observed in males that underwent lobe surgery, similar to what was observed by Rhebergen et al. . Partially amputated lobes ranged in length from lacking at least their most distal tip to more reduced stumps whose length approximated the corresponding right lobe.
Adult dissection and imaging
Adults were dissected in water with forceps (Forceps Dumont #5, Fine Science Tool) inside a transparent 25 mm round dissection dish under a stereo-microscope (Zeiss Stemi 2000). Male genitalia were isolated by piercing the abdomen with the forceps between the genital arch and the A6 abdominal segment and thereby separating the genitalia from the abdomen. We also dissected the left anterior leg of each dissected male. Female spermathecae were recovered after opening the ventral abdomen with forceps and removal of the gut and ovaries. The spermathecae were isolated and separated according to left and right sides and immediately examined using a microscope (see below). All dissected tissues were stored in 200 µL storage solution (glycerol:acetate:ethanol, 1:1:3) at 4 °C.
For imaging, male genitalia were transferred into a dissection dish filled with pure glycerol and examined with a VHX2000 (Keyence) microscope equipped with a 100–1000× VH-Z100W (Keyence) zoom objective at 300–400 fold magnification. Genitalia were oriented to be visible in posterior view. The left and right lateral spines, as well as the dorsal edge of the genital arch were aligned to be visible in the same focal plane. In some preparations, the genital arch broke and lobes were aligned without adjusting the position of the dorsal genital arch. The experiment was discarded from analysis in cases where lobes could not be aligned. Male lobe lengths were measured on acquired images as the distance between the base of each lateral spine and the tip of each lobe (Fig. 1c, d) using ImageJ version 1.50d (https://imagej.nih.gov/ij). Male legs were put on a flat glycerol surface with the inner side of the tibia facing to the camera. Legs were imaged at 200 fold magnification with the VHX 2000 microscope (Keyence) as described above.
We prepared the female paired sperm storage organs (spermathecae) at 12–24 h after copulation (54 females from stock 15090–1698.01 and 24 females from stock 15090–1698.02, Additional file 2) and examined the presence of ejaculate. D. pachea has giant sperm , which makes direct sperm counts difficult. Therefore, we determined an apparent average spermathecae filling level per female based on visual inspection, similar to Jefferson . Female spermathecae were arranged on the bottom of the transparent plastic dissection dish, filled with water. Images were acquired using transmission light in lateral view with the Keyence VHX2000 microscope at 400–500 fold magnification. Ejaculate was directly visible inside the transparent spermathecae. Ejaculate filling levels were annotated for each spermatheca separately to match three different categories: 0, 1/6 or 1/3 of its total volume. Then, the average filling level for both spermathecae was calculated for each female.
Annotation of courtship and copulation behaviour
Videos were analysed with the OpenShot Video Editor software, version 1.4.3 (https://www.openshot.org) to annotate the relative timing of courtship and copulation in our experiments (Additional file 1: Figure S2, Additional file 3). Data was manually entered into spreadsheets. We annotated the beginning of male courtship as the start of at least three consecutive male courtship behaviours according to Spieth , such as male touching the female abdomen with the forelegs, wing vibration, male following the female, and male licking behavior (see above). We estimated licking behavior throughout courtship of each male in trials were both males courted the female simultaneously. Licking occurs abundantly throughout courtship of D. pachea (Additional file 1: Figure S3) and was therefore chosen to assess the relative courtship contribution of the two males. We calculated the sum duration of all licking events per male and trial and the number of licking sequences. However, this behavior could not be quantified in 7 trials because the males changed positions very fast so that they could not be unambiguously distinguished (Additional files 3, 4). In all analysed trials, licking behaviour was observed throughout D. pachea courtship from shortly after courtship start to the start of copulation (Additional file 1: Figure S2). Both males reached at least once the licking position in all trials (Additional file 1: Table S3). This indicates that the female had physical contact with the proboscis of both males before copulation started (Additional file 1: Figure S2).
The beginning of copulation was defined as the moment when the male mounted the female abdomen. However, we only scored a copulation start if the male also settled into an invariant copulation posture for at least 3 s. Upon mounting, the male moves its abdomen tip forth and back along the female oviscape. This behavior lasts for up to about 3 min and during this time the male grasps the female wings and abdomen with its legs. This behavior was previously described as settling period and ends with a characteristic right-sided, invariant copulation posture [41, 42, 44]. Any copulation that ended within the first minutes after mounting without settling was considered to be a “failed mounting attempt” (Additional file 4). The end of copulation was considered as the moment when male and female genitalia were separated and the male had completely descended with the forelegs from the female abdomen.
Copulation duration (Additional file 1: Table S3) was comparable to previous analyses of D. pachea copulation durations [41, 44, 47, 57]. We observed that copulation ended upon isolation of the non-copulating male in 3 trials with females of stock 15090–1698.01 and in 4 trials with females of stock 15090–1698.02, probably indicating premature copulation end due to specimen handling. However, the estimates were included into the dataset and had little impact on average copulation duration when comparing mean and median values.
Our aim was to compare at least 50 mate-competition trials, where both males courted the female simultaneously and where copulation was observed. In total, we carried out 98 and 89 trials with females of stocks 15090–1698.01 and 15090–1698.02, respectively (Additional file 1: Table S2). We performed another 62 trials with males that underwent surgical manipulation of the left lobe and females of stock 15090–1698.01 (Additional file 1: Table S4). In total, we removed 36 trials from the analysis because either copulation was not observed until 1 h after recording started (21 trials), flies escaped, got injured or died inside the mating cell (4 trials) or the dissections of male genitalia failed (11 trials, Additional file 1: Table S4). We thus examined 152 trials with unmodified males and 61 trials with males that had undergone genital surgery, (see below, Additional file 1: Table S4). Among those, we observed that both males courted the female simultaneously in 111 and 54 trials, respectively. Statistical assessment with the Bradley Terry Model (see below) further required a complete dataset so that we excluded 20 additional trials with missing values, 19 trials with unmodified males (left lobe length: 2 trials, right lobe length: 2 trials, licking behavior and/or failed mounting attempts: 7 trials, tibia length: 8 trials) and one trial with lobe-modified males (no information on tibia length). In total, we considered 92 trials with unmodified males and 53 trials with lobe-modified males.
Data analysis was performed in R version 3.6 . Male copulation success (see above) was evaluated with a Bradley Terry Model  using the R package BradleyTerry2 . This model is suitable for logit fits to pairwise comparison data since it takes into account the special non-random (pairwise) structure of the dataset. We incorporated data (Additional file 3) into a list with three objects (Pacdata, available on DRYAD): (1) “contest” with the pairings of males, a trial ID and a factor indicating trials with males that had modified left genital lobes, (2) trial-specific variables “conditions” containing female age, female stock, courtship duration and a factor indicating which male initiated courtship, (3) male-specific predictor variables “male_predictors” with left and right genital lobe lengths, male tibia length, sum licking duration, number of licking sequences and failed mounting attempts. We developed the statistical model (Additional file 1: Tables S5–S11) by evaluating the effect of different variables on copulation success and by comparing pairs of nested models, using Chi-square tests with the generic anova() function . Outlier observations were considered when the standardized model residuals were greater than 2. The effect of corresponding trials on model fits was evaluated by comparison to additional model estimates without these outlier trials. We estimated correlations of male specific predictor variables and among numeric trial specific variables with the corrplot R package (Additional file 1: Figure S6). We found a significant correlation of the male predictor variables ‘sum licking duration’ and ‘number of licking sequences’. In addition, the trial specific variables ‘female age’ and ‘courtship duration’ were significantly correlated (Additional file 1: Figure S4). Correlated parameters were not incorporated into the same model for assessing the best fitting model (Additional file 1: Tables S5–S12). We tested trial specific variables in interaction with all male-specific variables, but only found an interaction of female age with failed mounting attempts in the dataset with unmodified males (Additional file 1: Figure S4, Table S8, Table 1). We also tested for interactions between male specific predictor variables but did not find any. The best fitting model for the datasets was: copulation success ~ sum licking duration + left_lobe length.
Spermathecae filling levels were evaluated with a linear model using the lm() function: spermatheca filling level ~ left lobe length + right lobe length + sum licking duration + Number of licking sequences + copulation duration + female age + male age, and significance of each variable was evaluated with an F-test using anova().