Exaggerated ornaments provide opportunities for understanding how selection can operate at different levels to shape the evolution of a trait. While these features aid their bearer in attracting mates or fending off competitors, they can also be costly and influenced by the environment and genetic variation. Eyestalks of the stalk-eyed fly, Teleopsis dalmanni, are of interest because eyestalk length is the target of both intra-and intersexual selection and is also reduced by loci on a highly-diverged sex ratio X chromosome (X^SR), a meiotic driver accounting for up to 30% of wild X chromosomes. Male stalk-eyed flies fight to control access to females and over food using a combination of low intensity displays and high intensity physical fights. We staged, filmed and scored contests among eyespan-matched male pairs to evaluate whether X chromosome type impacts the behavior and outcome of aggressive interactions. While our results broadly match expectations from previous studies, we find that X^SR males used more high intensity behaviors than males carrying a non-driving, standard X chromosome (X^ST), in particular when their eyestalks were of similar size or smaller than their opponents. Additionally, we find that when X^SR males use high intensity behaviors they win more bouts than when they use low intensity behaviors. Taken together these results suggest that X^SR impacts male aggressive behavior to compensate for the shorter eyestalks of X^SR males, and may help to explain how this selfish chromosome is maintained.

Following previous studies of fighting in stalk-eyed flies (Panhuis and Wilkinson 1999; Egge et al. 2011; Bubak et al. 2014), we quantified male aggressive behavior in two sets of trials conducted either at SUNY Geneseo (outbred pair trials) or the University of Maryland, College Park (standardized opponent trials). We conducted trials with outbred and inbred flies as opponents to confirm that the results were robust to the opponents’ genetic background. Sexually mature males at least four weeks post-eclosion were anesthetized using carbon dioxide, eyespan and body length were measured from video-captured images to the nearest 0.01 mm using ImageJ. All trials were conducted in transparent plastic arenas (10 x 3.5 x 6 cm) lined with moist blotting paper and initially divided into three sections by opaque plastic barriers.  Each male was placed into an outer compartment of the arena and fasted 20-24 h.  At the beginning of the trial, a piece of agar gel made with corn juice was placed into the center compartment as a food resource and the barriers were removed. Trials were video recorded for ten minutes. Scoring of each video was done blind to genotype. Occurrence of aggressive behaviors performed by each fly in each trial were scored from video recordings following a published ethogram. Within a trial, we defined fighting “bouts” to begin once either fly performed an aggressive behavior and end when the flies were more than a body length apart or not facing each other for 3 s without exhibiting any aggressive behavior. A male was scored as losing a bout if he moved away from his opponent at the end of a bout either slowly (away) or quickly (retreat) and his opponent did not. A mutual away or retreat was considered a tied bout. Most trials involved multiple bouts. Aggressive behaviors were categorized as either high intensity (HI) behaviors that involved physical contact (tussle, attack/lunge, jump attack) or low intensity (LI) behaviors that involved males mutually displaying their eyestalks in close proximity (approach, flex, and extend, line-up eyestalks). Escalations were tallied for each individual as a transition from a LI to a HI behavior with no other behaviors or the start or end of a bout separating them.  Similarly, de-escalations were defined as a transition from HI to LI behavior. Males were collected immediately after each trial for DNA extraction and genotyping for X chromosome type (X^SR or X^ST) using polymerase chain reaction (PCR) to amplify either of two markers (Table S1) that each predict drive phenotype in this population with 95% accuracy.