In this and the next two chapters we advise on the wise use of graphic formats, recommend several software tools, introduce a framework to help improve your graphics, and provide guidance on designing maps and data plots, including video tutorials to help you draft beautiful figures. If you are using this guide for the first time, start with the section on graphic formats, followed by those on software and a graphics framework. You could then move on to either Map with a message or Plot with a purpose, depending on your needs.
Figure 15: The difference between raster and vector. The raster image is composed of a fixed set of pixels, and the vector image is composed of a fixed set of shapes. Scaling the raster image reveals the pixels whereas scaling the vector image preserves the shapes. (By Yug, modifications by 3247 (Unknown) [CC BY-SA 2.5], via Wikimedia Commons.)
To be able to choose the correct graphic format for a figure—whether it is a map, data plot or photograph—it is important to understand there are two graphic types: vector, which consists of lines and other shapes, and raster, which consists of pixels (sometimes referred to as dots; Fig. 15).
A vector graphic uses points, lines, curves and other shapes or polygons to represent an image. Vector graphics have four particular advantages compared to raster graphics:
File sizes are smaller
An image can be rescaled to any size without loss of quality
Any included text can be read and therefore indexed by internet search engines
When rasterized for printing a vector graphic is equally sharp at all sizes
Figures such as maps and data plots are best prepared in vector format. The standard for this format is scalable vector graphics (SVG).46 Figures can also be prepared in postscript, a vector format that was originally designed for desktop publishing, although in our experience SVG produces more consistent results.
Figure 16: La récolte des pommes à Éragny, by Camille Pissarro [Public domain], via Wikimedia Commons.
A digital photograph is typically in raster format, as is the on-screen rendering of text and pictures in this guide, and the paintings of the Pointillism school of art (Fig. 16), in which dots of colour are applied to form an image.
There is a wide variety of raster formats. The images in the online version of this guide are in JPEG (Joint Photographic Experts Group) format, often abbreviated to JPG. This format is commonly used for displaying a graphic on a screen, as it allows an image to be compressed in size (and thus display quickly) yet maintains a reasonable image quality.
Publication-quality raster graphics (typically photographs, which are often referred to as plates) should be presented in TIFF (tagged image file format), which stores image data in a lossless way. This means that, unlike images in JPEG format, a TIFF image can be edited and re-saved without losing quality. TIFF images may be very large but can be compressed without loss of quality, and can include tags for geographical location.47 geoTIFF
TIFF files need to be provided at a resolution of 600 pixels per inch (commonly referred to as PPI48 Although often used interchangeably, the term pixels per inch should not be confused with dots per inch (commonly referred to as DPI), which is used to describe the number of dots per inch in a print from a digital file.) to achieve clarity and sharpness.49 Use of inch as a measure is of course outdated, but is retained for historical reasons associated with printing.
Journals normally specify the width (in mm or inches) at which they require figures and plates. Multiplying this width by the required pixel density gives the minimum width in pixels at which you need to supply a raster-format graphic. Table 10 shows the required sizes for three typical figure widths in mm.
Table 10: Minimum widths in pixels for a graphic in raster format (0.03937 converts mm into inches).
Width (mm) | Intended for | Minimum width required (pixels) |
---|---|---|
82 | Single column | 1,937 (82 * 0.03937 * 600) |
115 | 2/3 page width | 2,717 (115 * 0.03937 * 600) |
171 | Full page width | 4,040 (171 * 0.03937 * 600) |
There is a plethora of software available for drafting maps and plotting data. For ease of use and convenience in instruction, and to ensure that all readers have the same opportunity to follow the instructional videos on preparing publication-ready figures, the examples in this guide utilize only Free and Open Source Software that is available for all three of the commonly used operating systems (Windows, Mac and Linux). All the illustrations in this guide were produced with these tools.
For mapping we use the widely used geographical information system QGIS. It is maintained by an active user group and has a helpful mailing list.
For data plots we use both Veusz, which is maintained by a helpful designer and also has an active and responsive mailing list, and ggplot2, for which extensive help is available. Veusz is mouse driven whereas ggplot2 is command-line driven.
One of the strengths of QGIS, Veusz and ggplot2 is that, used correctly, they produce excellent publication-quality graphics.
Two other software tools are indispensable for producing high-quality graphics for publication: editors for raster and vector format graphics. Such editors are particularly useful for touching up and finalizing graphics following export from QGIS, Veusz, ggplot2 or other software, for converting between file formats, and for rescaling images. For raster graphics we recommend Gimp, and for vector graphics Inkscape. Gimp was used extensively to prepare grescale vesions of the figures in this guide (e.g. Figure 18) and Inkscape was used to edit SVG versions of some of the figures (e.g. Figure 69)
Here we introduce a framework of eight design principles for drafting clear, uncluttered graphics, following the cartographic design principles promoted by the UK Ordnance Survey. Although originally designed for drafting maps, the framework can be used to guide the design of any type of graphic, including data plots. The framework’s eight principles are paraphrased below and illustrated using a map figure from Hoffmann et al. (2015) (Figure 17) and a data plot from Cunningham et al. (2016) (Figure 19).
Figure 17: Locations where tree seeds were collected in fragments of the Araucaria moist forest in southern Brazil, and the original extent of this forest type (a modified version of Figure 1 in Hoffmann et al. (2015)).
1. Understanding user requirements—Communicate the message clearly
It is clear that the seed collection sites lie within the original extent of the Araucaria moist forest,50 A coniferous forest ecoregion of the Atlantic Forest Biome of southern Brazil and north-eastern Argentina, of particular conservation and evolutionary importance where this lies with respect to the southern Brazilian states and neighbouring countries (main map), and where southern Brazil lies in South America (inset map). The location of the main map is indicated by a self-explanatory rectangle on the inset map. International borders are included on both the main and inset maps.
2. Consideration of display format—Design for the medium
The figure has been designed at a width of 82 mm, for the single column width of the chosen journal.
3. A clear visual hierarchy—Draw attention to elements of interest and push those of less importance down the visual plane
Font weight, size and colour are used to draw attention to the elements of principal interest (Table 11). With regards to the labelling, the main message has the largest and heaviest font, state names are in a smaller and regular font, and the names of surrounding countries and ocean label have the smallest and lightest font. The inset label has a font size and weight that places it intermediate in the hierarchy. For the other elements, the legend is in a slightly larger and heavier font than the scale bar and grid tick labels. The symbol used for the collection sites has a diameter of 1.8 mm. The frames of the main and inset maps and legend, and the grid ticks, have a thickness of 0.2 mm. The international and state boundaries have the same line thickness (0.1 mm) but differ in shade, to separate them. The coastline has a thickness 0.1 mm as it is a complex line: being thin ensures that neighbouring lines do not run together in a visually displeasing manner. In general line thicknesses of 0.1–0.3 mm and fonts of 6–12 points are suitable for publication-quality figures.
Table 11: Hierarchy of font type, weight and size used in Figure 17. The font is Assistant.
Level | Element | Font weight | Font size (points) | Example |
---|---|---|---|---|
Labelling | ||||
1 | Main message | Semi-bold | 11 | Araucaria moist forest |
2 | State | Regular | 8 | Santa Catarina |
3 | Inset label | Regular light | 7 | Brazil |
4 | Neighbouring countries | Regular extra light | 7 | Paraguay |
4 | Ocean | Regular extra light italic | 7 | South Atlantic Ocean |
Other elements | ||||
1 | Legend | Regular | 7 | Collection sites |
2 | Scale bar | Regular light | 6 | 100 km |
2 | Grid ticks | Regular light | 6 | 52°W |
4. Simplicity—Include only the necessary information
Only the information required to convey the spatial information is included; there is no clutter and there are no confusing extraneous elements. Latitude/longitude tick marks are labelled only on the top and right borders (the left or bottom borders could also have been used but on this map there is less room on these borders), and lie inside the bounding box rather than outside (the latter would result in the area of interest being smaller within the overall design). There is neither north arrow nor compass (the latitude/longitude tick marks orient the map sufficiently) and no unnecessary arrow or lines connecting the inset to the main map (the rectangle on the inset clearly shows the location of the main map). The scale bar has simple divisions, and the units are a simple multiple of 10; it does not require a bounding box as it is clearly visible and understandable without one.
5. Legibility—Ensure all elements are readable, understandable and recognizable, and sufficiently large and clear
All elements are large enough to be legible at the intended display width of 82 mm. The white-filled circle symbols51 In the hierarchy of symbols that could be used here, a circle is the first choice and the text are clearly visible against the background. Some of the study sites are close together and the size of the circles minimizes overlap whilst attempting, as far as possible, to make them all visible. A sans serif font—which provides a clean look—is used for all labelling.
6. Consistency—Ensure balance, organizing features into groups
The names of the four states have the same font size and weight and are positioned as centrally as possible within the visible parts of the respective state so that they can be clearly identified as state names but do not obscure elements of principal interest. The ocean is labelled in an italic font to identify it as a different type of item compared to the terrestrial elements.
Figure 18: Greyscale version of Figure 17.
7. Accessibility—Consider format and intuitiveness
As this figure is intended to be in colour in the online version of the article and greyscale in print, colours were chosen so that a greyscale version (Figure 18) is interpretable without modification. The inclusion of a self-explanatory caption and legend ensures the information is accessible.
8. Good composition—Position all visual elements appropriately, for a clear and complete understanding
The inset map obscures only a small portion of the forest, and the legend and scale bar do not obscure any items of importance. The legend and inset are left aligned and positioned equidistant from the bounding box of the main map. Positioning the two inset boxes and the scale bar towards the edges ensures they do not obscure or detract attention from the principal elements. The letters and words of some of the labels have been spaced, to emphasize the extent of the elements to which they are referring. The forest label lies entirely within the forest, and is—as a hint—in green.
As with a map, a well designed data plot draws attention to the items of interest rather than to peripheral information, and conveys its purpose in a clear, simple and informative way. The framework of eight design principles can also be used to guide the design of a data plot, as paraphrased and illustrated here (Fig. 19).
Figure 19: The numbers of (a) newly licensed Chinese giant salamander farms and (b) salamander hatchlings produced during 2004–2012 in Shaanxi Province, China (a modified version of Figure 2 in Cunningham et al. (2016)).
1. Understanding user requirements—Communicate the purpose clearly The reader can clearly see, without distractions, that the number of licensed farms and the number of salamander hatchlings increased over time. The two variables (licences and hatchlings) have been plotted separately rather than within the same figure, as the latter would potentially be confusing and/or misleading (implying, for example, a relationship between particular quantities of the two variables, depending on the scaling of the y-axes).
2. Consideration of display format—Design for the medium The plot is designed at a width of 82 mm, for single column width in the chosen journal.
3. A clear visual hierarchy—Draw attention to elements of interest and push those of less importance down the visual plane A plot will usually have a less complex hierarchy than a map. For simple plots the same font size may be used for most or all lettering. In this case all labelling is in a 10 point font except for the axis tick mark labels, which are in an 8 point font, to accommodate the numbers. None of the text is in bold or italics (such emphasis would be an unnecessary distraction). Figure borders, tick marks and the data line have a thickness of 0.2 mm, the y-axis gridlines have a thickness of 0.1 mm and are in grey (i.e. they are lower down in the hierarchy), and the data points have a diameter of 0.2 mm.
4. Simplicity—Include only the necessary information Only the information required to convey the purpose is included; there is no clutter and no confusing extraneous elements.
5. Legibility—Ensure all elements are readable, understandable and recognizable, and sufficiently large and clear All elements are large enough to be legible at the intended display width of 82 mm. The black-filled circle symbols and text are clearly visible against the background. A sans serif font—which provides a clean look—is used for all labelling.
6. Consistency—Ensure balance, organizing features into groups A simple plot may have only one or two groups. Here the same symbols and lines are used for the two parts (it is not necessary to use different symbol and line types, as plotting the data separately indicates there are two variables).
7. Accessibility—Consider format and intuitiveness Use of colour in this plot would be redundant and therefore there is no need to prepare separate colour and greyscale versions for online and print, respectively. Inclusion of a self-explanatory, descriptive caption ensures the information is accessible.
8. Good composition—Position all visual elements appropriately, for a clear and complete understanding Axis labels are centred on their respective axes, and tick labels are centred on their respective tick marks. The two parts are aligned vertically, with a small horizontal space separating them. As the x-axes of the two parts are identical it is not necessary to label the tick marks in (a).
Page built: 2019-07-12