Authorship of scientific papers matters because it’s a way of claiming credit and assigning responsibility. The authors are all and only those people who are entitled to claim credit for the work reported in the paper. And they’re responsible for the work reported in the paper, and so (for instance) are the ones who get blamed if the paper turns out to be flawed and needs to be corrected or retracted. Further, order of authorship traditionally provides information about the relative contributions of the authors–degree of credit and responsibility, if you like. So, how do you decide who the authors are on your papers, and in what order they’re listed?
My own approach, which I think used to be fairly standard in ecology and is still pretty common, is to think of a paper as arising from three main activities: conceiving and designing the study, conducting the study (e.g., collecting and analyzing the data), and writing the paper. You’re an author if you make a substantive contribution to at least two of those three. You’re the first author if your overall contribution exceeds that of any other author (typically, because you’ve made a substantial contribution to all three activities). And other authors are listed in decreasing order of their overall contribution.
My approach has some implications. It means people who contribute only to data collection, such as research assistants or technicians, wouldn’t be listed as authors (they would be listed in the Acknowledgments, of course). It means that the PI shouldn’t be listed as an author just by virtue of being the PI, or even if they came up with the basic idea for the project or gave some suggestions on study design but didn’t make any other contribution.
As sketched above, my approach obviously doesn’t cover all possible situations. It’s particularly ill-suited to big collaborations in which each of the “main” activities is subdivided among many people. But it at least provides a starting point for thinking about other situations. For instance, contrary to what I just said above, I do think it would be weird for someone not to be an author if they made a substantive contribution to the writing, even if they didn’t do anything else. But that situation rarely comes up in ecology (it comes up in medicine, where hiring “ghost authors” who write the paper but aren’t listed as authors, isn’t unheard of). Similarly, I could see making someone an author if they were largely or entirely responsible for developing and designing the project, even if they didn’t contribute much in other ways. As a grad student, I did a project that my supervisor had designed as part of a grant he got before I joined the lab. He handed the project to me to do, as a way to get the project done and as a training exercise for me. Not yet knowing much about the topic or the study system (that’s why I needed the training!), I followed his design and proposed analyses to the letter. I also wrote the paper. And then I had to talk him into being second author, as I didn’t feel comfortable being sole author without having made any contribution to the study design.
Underpinning all of this is the intuition that authorship should go to the people who make contributions of their own to the project, as opposed to people who contribute only by following instructions (as with research assistants who collect data as instructed but make no other contribution). On the other hand, data sharing is increasingly common and important in ecology. So it’s fairly common for people whose only contribution was to collect the data (or instruct others to collect the data) to expect co-authorship of any paper using their data. Especially when those data come from ongoing long-term studies and monitoring programs, that need to justify their continued funding by showing that their data are leading to lots of papers. I’ll be interested to see how attitudes on this issue develop in future. If in future data sharing goes from being valued to being expected or required, then it’s hard to see how sharing data should entitle you to claim co-authorship (i.e. credit and responsibility) for papers produced by others using data you collected. For instance, if you deposit gene sequences on GenBank (which journals, funding agencies, and professional norms require geneticists to do), you aren’t entitled to claim co-authorship of any papers that later use those sequences.
There are always judgment calls involved in determining authorship. For instance, if I just make some comments on a draft ms written by someone else, does that count as a “substantive” contribution to the writing? (I’d say it depends how numerous the comments were and how much the ms was altered by incorporating them) As with all judgment calls, I think the way you learn to make them is to make them, and to talk about ones others have made. General principles only get you so far; there’s no substitute for thinking about specific cases.
The above is merely my approach to authorship. Different people have different approaches, which is a bit of a problem since others can’t interpret authorship order properly if they don’t know how it was determined. Many cellular and molecular fields are pretty much the opposite of ecology: the last author, not the first author, is considered to be most important. The last author is usually the PI. Indeed, in many molecular labs the PI is always the last author on every paper produced by the lab, presumably on the view that the PI is ultimately the one responsible for all of the lab’s work. I hear from colleagues that this approach to authorship is starting to creep into ecology, which dismays me. It’s totally foreign to how I was trained, and it smacks of allowing external incentives to determine authorship. On the other hand, as someone who (like everyone) has strong incentives to have my name on as many papers as possible, I admit that I feel tempted to take a more expansive attitude towards authorship than my supervisor took. I do wonder if the average number of authors per paper is going up in all fields of science not just because we’re all becoming more collaborative, but because we’re all taking a more expansive attitude and lowering the bar on authorship, so that we can all put our names on more papers.
Perhaps the way to deal with this is to get away from authorship order as a means of apportioning and recording credit and responsibility. Authorship order was fine as a summary of authors’ overall contributions when few papers had more than 2-3 authors, but times have changed. I think it’s great that some journals now require statements of author contributions, which are published as part of the paper (e.g., “John Doe designed the study, John and Jane Doe performed the experiments and analyzed the data, and Jane Doe wrote the paper with assistance from John Doe.”) I’ve sometimes included such statements in the Acknowledgments even when not required to do so, and I plan to always do so in future. I’d encourage others to do the same.
The other universal piece of advice I can give is that everyone involved in a project should talk about how authorship of any future papers will be decided as early as possible, ideally before the project even begins. That means, for instance, that supervisors should talk about authorship with their grad students before the students even start designing their research projects, and collaborators should talk about authorship before they actually begin working. I actually talk to prospective students about my general approach to authorship issues when they visit my lab. Far better to address the issue early on, at least in a broad way, than to have to deal with disagreements and misunderstandings later. Similarly, all authors should discuss and approve a statement of author contributions as early on as feasible.