Build a story. First you need to center the theme or the problem. In which big ecological idea you are going to place the paper. Second explain what we know, and what we don't know yet. Explain WHY we didn't know it yet. Third, explain what is your contribution to this knowledge gap (can be small or big, but set up the expectation) and what will you do. Do not try to solve too many problems, and focus on the knowledge gap that your data can cover. Finally, start all paragraphs with the important take home message and then explain the details. Do not mix ideas within a single paragraph. It's key to guide the reader and differentiate between the key messages and the secondary results, because most readers will read it fast.
Do not try to be sequential (intro, methods, results, discussion). Be iterative. I start by writing easy paragraphs belonging to methods, annotating key references, and then I focus on the key figures (one figure per key result, ideally three results/figures per paper). This figures can be sketches or qucik R outputs. Once you know your story, make an expanded abstract. BUT do not start writing until the story is in your head. This extended abstract forces you to see if the story flows. You also discover which results are trivial (not even mentioned in abstract), and for which claims you don't have a result to back them up (you need extra analysis!). Then use this abstract to expand the intro, complete methods, write the results in text. This can take several iterations. I follow the mantra "write drunk, revise sober" in a metaphoric way. The first pass is to lay down ideas, just type stuff as it comes to your mind and don't revise much. Don't care about grammar at this moment and even use Spanglish if this helps writing your story faster and avoids breaking the flow of ideas. Then stop, think, re-order paragraphs and start the revising of the text, looking for a good grammar, etc... At some point you add the first stab of a discussion and keep it iterative. The last iterations is about making the paper perfect. Unifying terminology, ensuring discussion answer all questions raised in the intro (and viceversa), polishing figures, adding bibliography, etc...
- How to Write a Scientific Paper: An Academic Self-help Guide for PhD Students by researcher Jari Saramäki. Watch the talk or the slides.
- The elements of style is not focused on academic writing, but it can be useful for non-native English speakers like myself.
- Despite the flashy title, Writing Science: How to Write Papers That Get Cited and Proposals That Get Funded contains solid advice and knowledge about academic writing with very pertinent examples.
- Daphne Gray-Grant, a publication coach, runs a writing blog with interesting tips on academic writing, from dealing with writer's block to mindmapping.