Sexual selection through female choice can lead to the evolution of multiple exaggerated sexual traits in males. Satin bowerbirds (Ptilonorhynchus violaceus) are an ideal species for investigating female choice based on multiple signals. Male satin bowerbirds court females with complex displays that include dance, vocalisations, mimicry, brilliant plumage colouration and bowers built from sticks upon a display court decorated with blue feathers, yellow leaves, snail shells, flowers and blue artificial objects. Females visit several males before deciding upon a mate and appear to have strong preferences for males with well constructed and highly decorated bowers. In this study I consider several components of male display simultaneously and develop a framework for investigating female choice by satin bowerbirds in the context of multiple signals. In addition, I document some of the basic biology of females because this is critical to understanding their responses to multiple signals.
To understand the process of female choice of mates, it is important to determine the potential number of males females may be able to sample and to investigate how female mate search strategies may influence male mating skew. I radio-tracked female satin bowerbirds to determine their home range sizes and the potential number of males females may have to choose from. In the Bunya Mountains there was a significant difference in the mean size of females' home ranges between years, and this in turn influenced the numbers of bowers within females' home ranges. In the 1997/98 season, female satin bowerbirds had an average of 6.63 bowers (n = 8; SE = 1.01 bowers) within their home range. In the 1998/99 season this figure dropped to 1.86 bowers (n = 7; SE = 0.55). However, this difference may be explained by a later radio tracking start date in the 1998/99 season which likely missed the first stage of female mate searching where females sample more males relative to the second stage of mate searching. When the start dates of the two years were equalised by removing fixes prior to the 15th October 1997 the difference between the two years in mean female home range size was no longer significant. This suggests that female mate searching may influence the size of females' home ranges.
I then used Monte Carlo simulations to model the mate search patterns of females. Female satin bowerbirds are known to have complex mate search patterns and to use experience and knowledge from past years when sampling males. The models showed that, unlike classical lekking species, male mating skew increased with the numbers of males that females sampled. Even though female satin bowerbirds did not sample many males in a single season, the effect of their cumulative mate sampling over several seasons, which effectively increased the number of males sampled, ensured that there was a skew in the distribution of male matings.
Relative to other bowerbird species, the correlates of male mating success in satin bowerbirds have been extensively investigated. However, few studies have looked at multiple traits simultaneously. I used statistical models (Restricted Maximum Likelihood Mixed Models) to simultaneously explore multiple display traits and determine which traits females had preferences for. based on video monitoring of male mating success. I found that the signals females used to choose bowers to visit differed from the signals used when copulation partners were assessed. The best positive predictors of female visitation rate were the numbers of white natural decorations, the size of the bower owner and the rate at which males performed solitary displays. In contrast, the strongest positive predictor of male copulation success was the painting rate of bower owners. Higher numbers of non-blue artificial decorations and an increased rate of bower destructions by rival males had negative effects on male copulation success. The present study demonstrated that female satin bowerbirds used several signal modes and suggests that a previously uninvestigated mode of signalling in bowerbirds, bower paint, which contains volatile compounds and saliva, may function as a chemical signal. I concluded that when bower decorations, bower construction quality and bower painting are modelled together, males' painting rates may explain more of the variation in mating success than decorations and bower quality do. Thus, suggesting that behaviours may be stronger predictors of mating success than decorations and bower quality.
The colour of light in forests may influence the behaviour of animals. The lighting conditions during displays at bowers may add yet another layer of complexity to the signals of males. For example, males may locate their bowers in microhabitats that experience light environments that maximise the conspicuousness of their decorations and dance displays. In addition, female satin bowerbirds may prefer males with more conspicuous displays. In this study I tested the signal colour predictions of Endler (1993), which are the colours that animals should use if they are seeking to maximise the conspicuousness of their displays. I used a portable field spectroradiometer to measure the colour of the ambient light at bowers and random points in the habitat surrounding bowers. I found that satin bowerbirds locate their bowers in a microhabitat that experiences light environments that maximise the conspicuousness of their displays and thus they appear to conform to the signal colour predictions of Endler. In addition, the light environments at bowers were a non-random subset of the available light environments in the surrounding habitat. Given that woodland shade can only occur when the bower is shaded, males should time their behaviours to occur in the shade. Bower painting was consistently observed to occur in the shade more often than expected by chance.
In order to test the various models of multiple signal evolution, future studies need to use data on male mating success to determine females' preferences for particular traits, and then investigate whether these traits signal male quality. In this study I found that female bowerbirds do not appear to choose mates based on all available male traits, therefore such traits may be unreliable traits that do not honestly signal male quality. However, because some non-preferred traits were correlated with preferred traits, these non-preferred traits may be redundant signals. Further work on the evolution of multiple signals in bowerbirds would be a fruitful area of endeavour.