to test hypotheses explaining probability of occurrence of accessory, broken, and 23 malformed antlers. By employing a multinomial logistic regression, we showed that the 24 probability of occurrence of broken and malformed antlers increased in males older than 25 2 years. Probability of occurrence of accessory and, in young males, broken antlers was 26 higher in individuals with increased body weight. Occurrence of malformed antlers 27 decreased over the study period. Contrary to our prediction, we did not detect an effect of 28 distance to forest on the probability of malformed antlers occurring. We conclude that the 29 main premise of compensatory culling is not supported in roe deer. 30 KEY WORDS: Age, antler, body mass, Capreolus capreolus, habitat gradient, harvest, 31 malformation, trend, trophy. 32Antlers of most Cervidae are believed to have developed primarily as signals for female 33 mate choice (Clutton-Brock 1982, Bubenik 1990a. The signaling function of antlers has 34 been supported by showing positive correlation between conspicuous traits of antlers, 35 such as size (i.e., length and mass) and structural complexity and some fitness surrogates. 36Examples include correlation with lifetime reproductive output, sperm quality, and social 37 status in red deer (Cervus elaphus; Bartoš et al. 1988, Kruuk et al. 2002, Malo et al. 38 2005, body mass and immunological capacity in white-tailed deer (Odocoileus 39 virginianus; Harmel et al. 1989, Ditchkoff et al. 2001a, body mass in moose (Alces 40 alces; Solberg and Saether 1994), and body mass and territorial status in roe deer 41 (Capreolus capreolus; Pélabon and van Breukelen 1998, Hoem et al. 2007, Vanpé et al. 42 2007. 43Signaling function of antlers has also been scrutinized in the case of more subtle 44 structural characteristics, referred to as fluctuating and directional asymmetry. The 45 former and latter have been defined as small random and systematic differences, 46 respectively, in the shape between right and left antler. Fluctuating asymmetry may 47 reflect aspects of individual quality in some populations (e.g., Folstad et al. 1996, 48 Pélabon and van Breukelen 1998, Bowyer et al. 2001, Ditchkoff et al. 2001b, Mateos et 49 al. 2008 but not all (Kruuk et al. 2003, Bartoš et al. 2007 study, we describe occurrence of accessory, broken, and malformed antlers in a 66 population of roe deer males harvested in western Poland. We test the following 3 67 hypotheses on the correlates of occurrence of these antler anomalies: 68 1. Probability of occurrence of accessory antlers is higher in older than younger 69 males. This prediction stems from the fact that the risk of repeated injury 70 increases with time. Roe deer is among the species particularly responsive to 71 develop accessory antlers (Bubenik 1990b). 72 2. Risk of antler breakage, which is primarily caused by severe injury, is random 73 in respect to age. Hence, we test the prediction that probability of occurrence of 74 broken antlers does not differ between young and old males....