Landcare Research - Manaaki Whenua

Landcare-Research -Manaaki Whenua

Dogs Offer a New Approach to Weed Detection

Rusty and John looking for velvetleaf in fodder beet in Southland.

Rusty and John looking for velvetleaf in fodder beet in Southland.

Fiona Thomson, an ecologist with Landcare Research, has been investigating whether dogs could be used as a tool to detect new and emerging weeds in the landscape. Some weeds are easily located, especially when they are in flower, but others are more cryptic or difficult to find, such as some grasses and small or seedling plants. “Early detection is the key to preventing weed spread, and when visual methods fail, this is where dogs can potentially make a big contribution,” said Fiona.

Recently the Ministry for Primary Industries (MPI) funded a dog, Rusty (a collie heading dog, also used for search and rescue, owned by John Taylor of Invercargill), to be trained to detect a major emerging agricultural pest: velvetleaf (Abutilon theophrasti). In March Waikato Regional Council trialled using Rusty and John to detect velvetleaf at farms previously searched by people using visual methods. Over a week Rusty was able to find around 100 more plants, some as small as a 10 cent piece. Fiona joined Rusty, John and Environment Southland recently to survey nine Southland farms. No velvetleaf was found – great news for the farmers whose farms were searched. The team then went to the Manawatu–Wanganui region, where once again Rusty showed his value in finding velvetleaf plants.

With the support of Environment Canterbury, Fiona is training her own border-collie cross dog, Tahi, to detect Chilean needle grass (Nassella neesiana), having gained approval from MPI to store and carry this plant for (dog) education purposes. Tahi is currently undertaking preliminary detection and obedience training. “The first step was to figure out if dogs can distinguish Chilean needle grass from other plants, but preliminary results look promising.” However, Fiona is not too worried if Tahi can’t detect Chilean needle grass in the field. “A dog can potentially be trained to detect multiple weeds, so if Chilean needle grass is not a good candidate, there are plenty of other emerging weeds to test.”

“Using dogs’ powerful sense of smell could provide us with a new and more effective way to detect cryptic weed species in the environment such as Chilean needle grass,” said Fiona. Chilean needle grass is one of the most serious emerging weeds in the Canterbury region, threatening over 2 million hectares of land, but is currently only known to occur at 16 sites. Chilean needle grass is difficult to distinguish from other grasses until flowering time, at which point seeds have already been formed, making eradication all the more difficult. If dogs could identify infestations earlier this could make a huge difference.

“We know that dogs can do the job and that we need better tools to detect emerging weeds. Now we need to do some research to determine the advantages of using a trained dog compared with other techniques such as sending out trained staff, and how to optimise the approach. Basically we need to understand what tool, or combination of tools, will give us the best chance of detecting and eradicating emerging weeds,” Fiona explained.
In Australia, where dogs have been used successfully to detect invasive hawkweeds (Hieracium aurantiacum, H. praealtum and H. pilosella), time-to-detection models have been developed to quantify the weed detection rates of dogs. These are used as a basis to compare the efficiency of human searchers with dogs, as well as helping to fine-tune the search intensity expected. The models also help with budgeting, providing valuable information on the number of hours that human and dog-handler teams are fit to work each day.

The comparison between dogs and human detection rates is not straightforward and requires considerable effort to control for different variables. A detector dog relies on finding plant odours, and their ability to detect individual plants is influenced by things such as wind speed, wind direction and the presence of other ‘distracting’ scents. In contrast, humans rely on visual cues, and detection rates are influenced by the diversity of the vegetation and the terrain being searched.

But isn’t training individual dogs too expensive? “The most cost effective time to stop an invasive species is in the early stages of invasion when they are at low densities, but this is also when the weeds are hardest to detect,” said Fiona. So there could potentially be huge economic benefits if new incursions could be detected and dealt with early.

An unexpected bonus of detection dogs is increasing public awareness. Because detection dogs have public and social media appeal, they can also help raise awareness of weed species and the importance of early detection.

Funding is needed to support the exciting potential of using detector dogs to find weeds. If you can help, please contact Fiona.