to different conditions. However, sheer costs of flexibility cannot explain the polymorphism 22 created by personality variation. In a correlative approach, we here tested whether mate 23 choice might act as a major driving force maintaining personality variation in the 24 monogamous, biparental rainbow krib, Pelvicachromis pulcher. We personality-typed all 25 males and females for their boldness (activity under simulated predation risk) and allowed 26 females to choose between two males that differed in their boldness (behavioural level and 27 consistency). Prior to the choice, females were allowed to observe both males, expressing 28 their natural boldness towards a video animated natural predator. Both sexes showed 29 personality differences in boldness over the short-and long-term. Furthermore, when 30 removing side-biased females, we found a dis-assortative mating preference for the 31 behavioural level and an assortative preference for behavioural consistency in boldness. 32Such preference patterns might facilitate effective parental role allocation during offspring 33 care and/or provide genetic benefits. Our results suggest that sexual selection plays an 34 important role in the evolution of personality differences. 35 36 Keywords: anti-predator behaviour, assortative, behavioural compatibility, cichlid, mate 37 choice, Pelvicachromis pulcher, personality, risk-taking, sexual selection, side bias 38 39 3 Individuals have to cope with a wide array of environmental challenges. Therefore, 40 flexibility in the expression of behavioural responses towards different and changing 41 conditions should be favoured by selection ). Yet, individuals often show 42 considerable consistent between-individual differences in behaviour over time and/or 43 contexts (Boissy, 1995). Such personality differences are common throughout the animal 44 kingdom (reviewed in Gosling, 2001; Kralj-Fišer et al., 2014) and have been shown for 45 various behavioural traits, such as activity pattern, aggressiveness, exploratory tendencies, 46 boldness and fearfulness (reviewed in Dall et al., 2004; Gosling, 2001;. 47Personality traits are moderately heritable (Ariyomo, Carter, et al., 2013; Patrick et al., 48 2013;Reif et al., 2003;van Oers et al., 2005) and have fitness consequences (e.g. Ariyomo et 49 al., 2012; Dingemanse et al., 2005;Smith et al., 2008), suggesting they are not merely non-50 adaptive noise that surrounds an adaptive optimum (Wilson, 1998). Nevertheless, 51 underlying mechanisms that generate and maintain behavioural polymorphism are largely 52 unclear and many aspects of the growing body of theoretical frameworks still remain to be 53 empirically tested (reviewed in e.g. Schuett et al., 2010;Wolf et al., 2010). 54 55 Recently, Schuett et al. (2010) pointed out that sexual selection may be important in 56 generating and maintaining personality variation though this possibility has rarely been 57 tested (but see e.g. Montiglio et al., 2016;. According to the proposed 58 framework , personalities are expected to p...