Current Zoology 61 (4): 765–772, 2015 Received Apr. 24, 2014; accepted Oct. 29, 2014.  Corresponding author. E-mail: bibiana.rojas@jyu.fi. † These two authors contributed equally to the writing of this paper. © 2015 Current Zoology Frequency-dependent flight activity in the colour polymorphic wood tiger moth Bibiana ROJAS †* , Swanne P. GORDON † , Johanna MAPPES University of Jyvaskyla, Centre of Excellence in Biological Interactions, Department of Biology and Environmental Sciences, PO Box 35, FI 40001, Finland Abstract Predators efficiently learn to avoid one type of warning signal rather than several, making colour polymorphisms un- expected. Aposematic wood tiger moth males Parasemia plantaginis have either white or yellow hindwing coloration across Eu- rope. Previous studies indicate that yellow males are better defended from predators, while white males have a positively fre- quency-dependent mating advantage. However, the potential frequency-dependent behavioural differences in flight between the morphs, as well as the role of male-male interactions in inducing flying activity, have not been previously considered. We ran an outdoor cage experiment where proportions of both male morphs were manipulated to test whether flying activity was frequency- dependent and differed between morphs. The white morph was significantly more active than the yellow one across all treatments, and sustained activity for longer. Overall activity for both morphs was considerably lower in the yellow-biased environment, suggesting that higher proportions of yellow males in a population may lead to overall reduced flying activity. The activity of the yellow morph also followed a steeper, narrower curve than that of the white morph during peak female calling activity. We sug- gest that white males, with their presumably less costly defences, have more resources to invest in flight for predator escape and finding mates. Yellow males, which are better protected but less sexually selected, may instead compensate their lower flight ac- tivity by ‘flying smart’ during the peak female-calling periods. Thus, both morphs may be able to behaviourally balance the trade-off between warning signal selection and sexual selection. Our results emphasize the greater need to investigate animal be- haviour and colour polymorphisms in natural or semi-natural environments [Current Zoology 61 (4): 765–772, 2015]. Keywords Frequency-dependent selection, Flight, Colour polymorphism, Sexual selection, Aposematism Animals employ a variety of strategies to deter pre- dators (Endler, 1986). Aposematic animals advertise their inedibility through conspicuous colours, odours or sounds, which predators associate with chemical or physical secondary defences and subsequently learn to avoid (Poulton, 1890; Cott, 1940; Ruxton et al., 2004; Rojas et al., 2015). The effectiveness of these warning signals relies on their abundance (Lindström et al., 2001), as predators need a threshold number of encoun- ters with defended prey in order to create an aversion (Endler and Rojas, 2009). Several studies show that pre- dators are better at learning one type of warning signal than several and, thus, signal polymorphisms are not ex- pected (i.e. positive frequency-dependent survival se- lection) (Endler, 1988; Joron and Mallet, 1998; Endler and Mappes, 2004). Surprisingly, however, several taxa show intra-specific variation in warning signals (Myers and Daly, 1983; O'Donald and Majerus, 1984; Brake- field, 1985; Ueno et al., 1998; Mallet and Joron, 1999; Borer et al., 2010; Rojas and Endler, 2013; Hegna et al., 2015). Several findings have been proposed to explain co- lour polymorphisms in aposematic species, including: a combination of negative and positive frequency-depen- dent selection (Thompson, 1984); trade-offs between natural and sexual selection (Nokelainen et al., 2012; Crothers and Cummings, 2013; Cummings and Crothers, 2013); spatio-temporal variation in selection (Endler and Rojas, 2009; Valkonen et al., 2012; Galarza et al., 2014; Mappes et al., 2014; Nokelainen et al., 2014; Ro- jas et al., 2014b); an association between behaviour (or other fitness-related traits) and signal type (Nokelainen et al., 2013; Rojas et al., 2014a); relaxation of selection towards warning signals (Amézquita et al., 2013; Ri- chards-Zawacki et al., 2013); or non-adaptive forces such as hybridization or drift (Gray and McKinnon, 2007; Medina et al., 2013). Thus, it seems likely that several non-mutually exclusive mechanisms are respon- sible for the maintenance of warning signal variation, despite the expected strength of positive frequency-