Roses are red, violets are blue - so how much replication should you do? An assessment of variation in the colour of flowers and birds

After years of qualitative and subjective study, quantitative colour science is now enabling rapid measurement, analysis and comparison of colour traits. However, it has not been determined how many replicates one needs to accurately quantify a species' colours for studies aimed at broad cross-species trait comparison. We address this major methodological knowledge gap. We first quantified and assessed the variance in colour within and between species. Reflectance spectra of flowers from ten plant species and plumage of 20 bird species were measured using a spectrometer, and reflectance (i.e. brightness) and tetrahedral colour-space coordinates were calculated. analysis of variance (ANOVA) analyses indicate that there is far more variation in the colours of birds and flowers between species (>77%) than within species. A Mean Absolute Deviation from the Mean test was applied to indicate the sampling replication required for each species. Tetrahedral coordinates were sampled precisely with only one individual per species. Greater replication was needed to sample reflectance with the desired precision, particularly for darker coloured species. Our findings will allow researchers to allocate their sampling effort in a way that maximises the precision of their colour data collection. The fact that only a few replicates per species are necessary will greatly facilitate broad cross-species comparisons of colour in the future.