American Scientist

It is a rare research paper in which the authors are not inclined to note in the conclusions that “more research is needed.” This ubiquitous phrase sums up a central truth about science: It’s an ongoing process, and there is always more to learn.

But what should we do when a decision urgently needs to be made? For instance, habitats are being destroyed quickly, and the species that are struggling for survival in the remnants may not hang on until there is a truly robust data set on which to base determinations of what is the most efficient plan for conservation.

Two articles in this issue tackle this conundrum, through somewhat different focuses. In the cover story, “Connecting Habitats to Prevent Species Extinctions,” Stuart L. Pimm and Clinton N. Jenkins make the case that a significant portion of the most endangered species are the ones that had a small habitat range to begin with, so when its size decreases, there’s not enough space to support a viable population, and the animal has nowhere else to go. The authors point out that small-ranged species have often been the ones studied last, after more charismatic species, so we have little data on them, and many have yet to be identified. Although many more studies need to be conducted, say Pimm and Jenkins, the few that have been tracking areas since before habitat loss happened have shown how much more slowly the larger remaining fragments have lost populations. They have acted on this data by working to build corridors between fragments and effectively expand the available range. So far, it seems to be working, even if, ahem, more data are needed.

That more conservation resources are spent on charismatic species is a point echoed by Manu E. Saunders in this issue’s Perspective column, “No Simple Answers for Insect Conservation.” Saunders is reflecting on the situation of insects and some recent papers that have concluded that insects are undergoing worldwide decline, with media coverage of those reports claiming an “insect apocalypse” is imminent. Saunders explains that the available data actually show that insect populations are far more variable, and that they go through many boom and bust cycles that are poorly understood. She cautions against overextrapolation of data, and calls for more data collection on oft-maligned insects and their habitats.

In another account of the importance of checking and rechecking one’s work, Brian Hayes returns to the Computing Science column for this issue to celebrate the upcoming 50th anniversary of the Apollo program’s greatest accomplishment; that program is legendary for building in checking-and-rechecking procedures that NASA astronauts are still known for today. But Hayes isn’t as concerned with the astronauts who walked on the Moon as he is with the often-overlooked guidance computers that the modules were equipped with, and the intrepid programmers of those computers. Those computers controlled so much and had so many ways in which they could go wrong, but didn’t. In “Moonshot Computing,” software engineer Margaret Hamilton and many members on the Apollo Guidance Computer team tell Hayes that the robustness of the programming was the result of extensive, careful checking. (An extended interview with Hamilton is available.)

So please, go ahead and thoroughly collect your own observational data on this issue by giving these articles a read, and let us know what you think constitutes sufficient data collection and error checking in your area of science.—Fenella Saunders (@FenellaSaunders)