Carotenoid limitation and mate preference evolution: A test of the indicator hypothesis in guppies (Poecilia reticulata)

Under the indicator models of mate choice, female preferences evolve to exp loit the condition-dependence or "indicator value" of male traits, which in turn may cause these traits to evolve to elaborate extremes. If the indica tor value of a male trait changes, the payoff function of the female prefer ence for that trait should change and the preference should evolve to a new optimum. I tested this prediction in the guppy, Poecilia reticulata, a spe cies in which the indicator value of a sexually selected male trait, carote noid coloration, varies geographically. Carotenoid coloration is thought to be an indicator of foraging ability and health because animals must obtain carotenoid pigments from their diet. The primary dietary source of caroten oids for guppies is unicellular algae, the abundance of which varies among natural streams because of variation in forest canopy cover. Carotenoid ava ilability limits male coloration to a greater extent in streams with greate r forest canopy cover. Thus, the indicator value of male coloration covarie s positively with canopy cover. To test the indicator model prediction, I m easured genetic divergence in the strength of female preferences for carote noid coloration between high- and low-carotenoid availability streams in ea ch of three river drainages. Second-generation laboratory-born females were given a choice between full-sib males raised on three different dietary le vels of carotenoids. For all six populations, male attractiveness (as deter mined from the responses of females to male courtship displays) increased w ith dietary carotenoid levels. However, the strength of female preferences differed between populations in the predicted direction in only one of thre e river drainages. These results fail to support a crucial prediction of th e indicator model. More studies taking an interpopulation approach to study ing mate preference evolution are needed before the explanatory value of th e indicator models can be rigorously assessed.