If you don’t think figures are crucial to effective scientific communication, think about how much easier it is to remember what a bird looks like than to remember its song. We all know about the special stomach just for desserts. There is a special corner in our brains for figures. We remember research better with good figures. You want your research to be remembered, to be cited, and to get into the blogs and textbooks.
Scientific figures take two broad forms. In one, they present the data, using graphs, cladograms, gels, and the like. In the other, they present hypotheses showing linkages among concepts, or concepts and data. Both should be clear. You should aim to have both kinds in most papers and to present all the important data graphically.
This is the place where you need to be especially vigilant against lab jargon. Have you labeled your axes in ways that are intelligible even without reading the figure caption? Do it if you can. Think of the impatient reader that will read your abstract, then glance at the figures. Make them compelling enough to keep that reader interested.
Every journal has advice for exactly how to do your figures. Pay attention. Here are some sources of specific advice: here, here, and here. You could get some great books at the library too. You can look at the figures in papers you like and pay attention to their style. The point is, once you start focusing on this issue, you will find easy ways to improve your figures. It is important and worth the time.
There are a few things you should always do. First, make all the figures in one paper scrupulously identical in form. Keep the fonts the same. If you call something cats in one figure, don’t call it felines in the next, for example. Second, do not use the graphics defaults. This will result in too many significant digits on the axis label, too small numbers, and other similar issues. Third, make your figures beautiful and clear. They are your first contact with your reader.
There are lots of packages you can use for figures. Some kinds of data come from specific programs. But for basic figures generated from numbers you input, I recommend using R. It is open to all and is something you can keep using when you move out from whatever site licenses your current institution has. The current standard is GGPLOT2. The thing about R is whatever problem you have, someone else might have solved, as in this case.
Once you embrace the utility of figures, you will realize that they can be useful in teaching and before the research is done. A student should be able to plot many kinds of anticipated results with a clear graph. What are the independent variables? What is the response variable? What is the relation between the two?
I suppose it is crazy to write about figures with nary an illustration, but I am more than usually focused on paper writing.