Jason Sheltzer is a graduate student in cancer genomics at the Massachusetts Institute of Technology in Cambridge. He works in the Amon Lab, where the principal investigator (PI), half the graduate students and half the postdocs are women. Sheltzer was astonished when a friend at Princeton University told him she was the first female graduate student at her PI’s physics lab in his 20-plus years as an academic.

In fact, that bit of news prompted Sheltzer to take a closer look at gender distribution in academia in his discipline—biomedicine—which, in contrast to many scientific and science-related fields, isn’t completely dominated by men. That relative abundance of women, however, only makes the number of female faculty in tenure-track positions harder to explain: Slightly more than a third of all assistant professors are women, and the percentage of female full professors is about half of that. Some analysis was in order, he decided.

Along with his partner—Joan Smith, a software engineer at Twitter—Sheltzer gathered information on 4143 graduate students, 4904 postdoctoral fellows, and 2062 faculty members employed at 24 of the 25 highest-ranked U.S. biomedical research institutions. (They left out one—Scripps Research Institute—because that institution classifies postdocs in ambiguous ways.) The results of the analysis are described in the latest issue of the Proceedings of the National Academy of Sciences, in a paper titled: “Elite male faculty in the life sciences employ fewer women.” High-achieving male faculty, the paper concludes, train fewer women than other investigators do—fewer than elite women and fewer than non-elite men.

That matters, because working in an elite lab is one of the best ways for talented, academe-oriented scientists to take the next step into a tenure-track faculty post. In fact, the study found, assistant professors at elite institutions are largely drawn from postdocs at elite labs.

The study’s authors define as “elite” labs led by Howard Hughes Medical Institute (HHMI) investigators, members of the National Academy of Sciences, and winners of major career awards. The study shows that, on average, male faculty at top institutes employ fewer female graduate students and postdocs than female faculty at those same institutes do, and elite males—as defined above—train even fewer. Consequently, women are underrepresented as postdocs in these important feeder labs—surely an explanation for the paucity of women faculty at these top institutions.

The existence of a “leaky pipeline” for women in the life sciences is well established, says Nobel laureate Carol Greider of Johns Hopkins University School of Medicine in Baltimore, Maryland. (This problem—which is seen in all science, technology, engineering, and mathematics fields—was documented in the American Association of University Women report, “Why So Few?”) Nationally, although half the graduate students and a comparable number of postdocs in biology-related fields are women, only 36% of assistant professors are women. At the top research institutions surveyed, 29% of assistant professors are women, and that number shrinks at each professional promotion step. “This study would suggest that it is not just that fewer women choose to go from postdoc to assistant professor, but that there might be less of an opportunity for them to actually compete for such positions,” Greider says. “Instead of asking, ‘Why are there fewer assistant professors?’ we can now ask, ‘Why are there fewer women in the competitive labs that feed the candidate pool?’”

This gender imbalance comes as news to some elite male PIs contacted by Science Careers. “I was unaware of this very surprising bias myself until I read this paper, which I believe makes a very important contribution to keeping our community of scientists maximally healthy and productive,” says Bruce Alberts of the University of California, San Francisco, via e-mail. (Alberts is also a former editor-in-chief of Science.) “The only real way to help ameliorate this problem is to make us all aware of it, so that we try harder to protect ourselves from unconscious biases that are likely to be a substantial contributor to this data, in my opinion.” Randy Schekman, an HHMI investigator and Nobel laureate, finds the numbers discouraging but echoes the study itself in cautioning that the findings don’t necessarily indicate bias on the part of male PIs. Self-selection by female students and postdocs may also play a role: Women may prefer working in labs headed by women.

Greider, however, does not buy the self-selection argument. If these women choose to go to certain labs, they are not making such choices in a vacuum, she says. There is a pervading cultural problem in top labs that needs to be addressed by men and women alike, she adds. Greider is referring to the “unsustainable hypercompetitive system” that Alberts and co-authors highlighted in their recent PNAS article, “Rescuing US biomedical research from its systemic flaws.”

This survey, Sheltzer says, is a snapshot of the field of biology at a particular instance. “Labs (even among the elite) vary greatly in the number of women hired.  I absolutely don't believe that every lab, which is, say, 80% men on the day we surveyed, has a sexist PI or a problematic culture.”

The authors do not, however, rule out that a hostile (or unfriendly) climate could be repelling women from some of those labs. “A lack of applications from women could also reflect specific issues with a laboratory or PI,” the authors write in the article. “[F]aculty members who are reputed to be hostile toward maternity considerations could be implicitly discouraging women from applying to their laboratories. More insidiously, 16% of women employed in the academy report that they have experienced work-related sexual harassment, and the tolerance of sexual harassment in a laboratory could further decrease the number of female applicants.”

The best outcome of this paper, says Sheltzer, would be for male PIs who receive few applications from women to go out and find female candidates—at talks, conferences, and seminars—and encourage them to apply. As the paper points out, there is no dearth of well-qualified women.

Vijaysree Venkatraman is a Boston-based science journalist.

10.1126/science.caredit.a1400167