Skip to main content

Amateur naturalists improve scientific data on tree diseases

Tuesday 25 July 2017

A new study concludes that members of the public can accurately report disease outbreaks affecting our native trees and that by combining their findings with official survey effort the quality of predicted disease distributions is improved.

Published in Royal Society Proceedings B, the work comes from a joint team of scientists from Rothamsted Research, the University of Salford and Forest Research. The study was part of a programme to understand the causes and distribution of acute oak decline (AOD) in England and Wales. It developed and refined a mathematical model that enabled a formal survey to make the most of a catalogue of reports from the public about AOD affected trees.

Amateur naturalists can provide valuable scientific data by reporting observations in their gardens and beyond. Such assistance from “citizen scientists” could have a huge impact on the scope of future surveys, according to Nathan Brown, a computational biologist at Rothamsted, which led the study. “Sightings from the public provide a valuable resource.  The challenge for researchers is to make the most of their efforts by developing new ways to use it.” he says.

Matched locations

The results showed clear similarities between the locations of reports gathered from the scientific survey and those from the public. However, the greatest benefits were seen when the datasets were combined - Using all available information increased the accuracy of predicted distributions.

The work has proved so successful that the team is keen to work further with citizen scientists and is directing anyone interested in contributing data to use the tree alert website provided by Forestry Commission’s Tree Health Diagnostic and Advisory Service (THDAS), or to register online at Observatree, a collaborative network of volunteer tree health surveyors who are trained by Forest Research and managed by the Woodland Trust.

“Citizen science is an exciting way to boost numbers on the ground, but we also need a way to make sense of the numbers that come back,” says Stephen Parnell, an epidemiologist at Salford. “This paper showed that citizen science information can be successfully integrated into mathematical approaches to predict where disease is”.

The challenge when using reports from the public, even reports subsequently verified by experts, is the inherent bias of the sampling; Public reports are ad-hoc, sent in at times and places that are convenient. Statistical analysis assumes that sampling is random, so new methods are required.

Early warnings

For this study, the researchers drew on an existing survey of diseased orange groves in Florida where experts had recorded the condition of every one of a region’s 240 000 trees to produce a detailed map of the disease’s distribution. Simulations could then select small samples from this dataset to test the accuracy of the prediction method, a process which allowed the effects of including verified reports from the public to be quantified.

"We were really pleased with the accuracy of that final map” notes Brown. "Volunteers can provide an efficient early warning system for pests and diseases, and our findings suggest that their discoveries can be used to define the distribution of affected woodland” 

The results are welcomed by Sandra Denman, a pathologist at Forest Research. “Having a reliable account of the distribution of AOD is crucial. Researchers are then armed with critical information about spread, and can estimate the locations of vulnerable trees, which leads to a greater understanding of how the disease operates.”

The project received funding from Department for Environment, Food & Rural Affairs and the Forestry Commission. Research into the distribution of AOD continues with support from Woodland Heritage.