I’m a proud co-author of an upcoming manuscript led by Gurutzeta Guillera-Arroita on optimal surveillance for detecting pest species. (Available for preview here; Guru’s already written about it here.)
In a previous paper, Mick McCarthy and I optimised a trade-off between investing in survey effort to find an invasive weed, and the damage it could cause if it remained undetected. We also looked at spreading that effort over a landscape, where dispersal and habitat suitability might alter the probability of weed occurrence, varied terrain might affect detectability and we might be constrained to a surveillance budget. It’s a handy guide for prioritising places for weed surveys and can tell us when it’s good sense to give up on a site and move on, even when we haven’t found anything.
But surveys could stop early when we do find something too, since the find triggers management action. Any extra effort we were willing to use at this site could be deployed to other places. Guru has made this adjustment to the design. While this sounds small, it requires several pages of calculus to straighten out (and the best bits are in the Supporting Information for your enjoyment).
It turns out that we can afford to allocate a little more effort to all sites, since a number of surveys will stop early. This gives us an extra boost of confidence that our non-detections correspond to true absences. This can make for more efficient surveillance in certain circumstances, particularly when the probability of weed occurrence and the benefits of early detection are high (see figure up top).
I’m chuffed that my earlier work served as some inspiration, and was thrilled to collaborate with people who dig Kuhn-Tucker conditions as much as I do. Here’s hoping it proves useful to some survey designers very soon. (Maybe we should give Phil a call.)
Guillera-Arroita G., Hauser C.E. & McCarthy M.A. (in press) Optimal surveillance strategy for invasive species management when surveys stop after detection. Ecology & Evolution.