When you want to use colour in your figures, always ask yourself first if there is a reason for it¹³. Some good reasons to use colour in your figures is to add a third axis to a 2D scatter plot, e.g., to indicate different categories in a scatterplot, or to indicate different features in a map. However, if adding colour will just add visual interest to your figures, keep your figures in black and grey tones as adding colours without a specific reason will only be confusing for your audience.

Do not use too many colours since this will distract from your message, and it is difficult for colour-blind people to distinguish between them. It is important that your colours are easy to distinguish from each other and the background.

If you do use colour, think about what colours to use. Default colours will often look unattractive and can make your figures look cluttered¹⁹. Moreover, our human perception of colour is complicated, and colours often have an intuitive meaning to people depending on the context²¹. For example, on a map, people will intuitively interpret blue as water, whereas in a temperature plot, blue intuitively indicates low temperatures and red indicates high temperatures. Take these connotations into account when you are creating a figure, and do not just use default colour settings of the software you are using as this may complicate your message. If you want to highlight certain features, it is best to use ‘pure colours’ to highlight important aspects in your figures²¹. For an example of intuitive colour use in a map, see Fig. 1 in this paper by Prebble et al. (2016).

About 8 to 10% of the population is colour blind²². Avoid green-red palettes as these are not distinguishable for visionally impaired people. More and more thoughtful colour palettes are available in colour packages for programming languages and other software. Pay extra attention to this when you are making your own colour palette. As an alternative, it’s possible to use different patterns instead of different colours. Be aware, however, that many patterns may unnecessarily clutter your graphs, and it may be sufficient to critically choose one of the many other available colour palettes.