Practicable methods for weed delimitation

A comparison of
In November 2011 Kate Giljohann & I recommended that Parks Victoria and the Department of Primary Industries search the places highlighted in map (a) for hawkweed infestations. They made it to the places in map (b) (purple tracks) and only found new infestations (red dots) close to the known population extent.

I’ve got a new article available for view online early! I’m proud to lead a team of researchers and weed managers discussing and demonstrating how we can manage weed delimitation.

‘Delimitation’ is a crucial process during a weed eradication program, whereby the full extent of the invasion is mapped out. There’s really no other way of ensuring that your removal method is successfully targeting every last pesky plant. In 2011 the Hawkweed Eradication Project Control Group were wondering whether they’d nailed it. They’d been sending out search teams for years and had drawn a minimum convex polygon (MCP, dashed line mapped above) around the infestations they’d found. But could there be more hawkweed beyond those boundaries? The worst case scenario was that hawkweeds had actually spread much further south, where the terrain gets so tough that they’d probably have to give up on eradication entirely.

‘Further south’ encompasses a huge area, and the Control Group consulted Kate Giljohann and I to prioritise locations for search. While working on a biosecurity surveillance review for ACERA the previous year, I’d learned that very little research literature directly addresses how we should design delimitation surveys! So Kate and I started afresh, adapting my previous hawkweed survey design to increase the value of hawkweeds detected outside the known population boundary. This process also accounts for hawkweed occurrence (via dispersal and habitat suitability) and detection rates, ensuring that search teams visit places where they’re most likely to find far-reaching hawkweed infestations.

The hawkweed project managers used the shaded map we supplied (a above) to direct their GPS-tracked search teams (b above). The good news is that they found no new hawkweeds in the dreaded southern reaches, just a few extra plants to the north and east of the infestations they already knew about. While this was good evidence that they were close to delimitation, the search teams’ GPS practices were inconsistent – it was hard work for Kate and Michael Rigby to clean up the spatial data, and we had little idea how much time and effort the search teams had put in at each site.

In our new article, my co-authors and I document the process of designing, implementing and evaluating this delimitation survey. I’ve even put together a wishlist for future delimitation work. While minimum convex polygons describe infestations that we know about, probability maps can tell us where else our weed population might be lurking. The maps are ripe for Bayesian updating as we collect new survey data. They can also be aggregated into a delimitation score developed by Mark Burgman, Dane Panetta and colleagues to track progress from survey to survey.

Based on what my co-authors and I achieved in this study, I’m optimistic that these measures can be taken in real eradication programs.


Hauser, C.E., Giljohann, K.M., Rigby, M., Herbert, K., Curran, I., Pascoe, C., Williams, N.S.G., Cousens, R.D. & Moore, J.L. (in press) Practicable methods for delimiting a plant invasion. Diversity & Distributions. DOI: 10.1111/ddi.12388