Most populations of weeds and pests already cover a large area by the time they are considered problematic. As a result, most invasive populations end up being managed by multiple people, each only having the legal right and/or economic incentive to manage the population on their land. Despite this, the vast majority of research on invasive species management assume that a manager can coordinate control strategies perfectly across the landscape.
In our paper we wanted to know how the collective decisions of many property managers affected the spread of two damaging weeds; serrated tussock and African lovegrass. I talk about the results of this work in the video below.
We looked at the spread of these two weeds at a large scale and over a longer period of time, in this case, 50 years. The only practical way to do this was to build a model. In our model there were 4096 decision makers who all decided whether or not to control the weed based on a simple rule: if the benefit of controlling a weed was higher than the cost, then that decision maker was more likely to control. If the reverse was true then they were less likely to control.
Using this model we could see how aspects of the decision maker’s behaviour, like how much each one cared about the weed, interacted with the ecology of the weed to affect the spread of serrated tussock and African lovegrass. Of course our model was not perfect and simplified many processes, but it was detailed enough to give us an idea of the effect that multiple land managers can have on weed spread. In addition, we were able to establish what aspects of their behaviour might be important.
Differences in the impact and ease of control between serrated tussock and African lovegrass highlighted an important point about which types of weeds are likely to become wide spread. Serrated tussock is very damaging to graziers, in some cases halving stock rates. But there are some practical steps land managers can take to reduce its density and impact. African lovegrass on the other hand is slightly less damaging but is very hard to control. This meant that the decision to control serrated tussock was very obvious, while the decision to control African love grass was more ambiguous. Consequently, most of our modelled decision makers controlled serrated tussock straight away and its spread was greatly reduced. African love grass on the other hand frequently took over the entire modelled landscape because there was always a pool of infested areas available to spread it. This suggests that weeds which go on to become wide spread, may be species with impacts that are not so large that everyone controls them, restricting their spread, nor so mild that they are not considered weeds.
This explanation does not tally with the reality that serrated tussock is widespread in Australia. However, a lot of serrated tussock spread happened before systematic efforts were targeted at controlling it. More recently there has been some success at reducing its prevalence in infested areas, largely because private landholders have participated in regionally coordinated control drives. This highlights another important finding of our model. For concern about a weed to have a large effect on its spread, that concern must be present while the weed is still rare (i.e. early in the invasion). If concern only grows as the species becomes widespread, then by the time there is enough concern that everybody acts, the weed is already so well established that landscape wide control is very difficult. Local government has an important role in this regard, publicising potentially damaging species before they become common.
A further complication when thinking about multiple actors is that not all of those deciding whether to control or not have the same goals or resources. They might be large commercial farms, smaller non-commercial farms or government organizations that look after the roads or parks. Some of these land managers may not even want to control invasive species at all. We included this in the model by making some decision makers need extra benefit to control weeds. We called these decision makers unmotivated and in our model they very rarely controlled the weed. Because unmotivated decision makers properties act as a constant source of the weed, they have an especially large effect on the rest of the landscape when there is lots of long distance dispersal (such as accidental movement of seeds by road). We found that a few unmotivated land managers (1 or 2%) didn’t make much difference. However, if they made up about 10% of all land managers, they could cause the weed (especially African love grass) to spread across the whole landscape. The reason why such a small change in the proportion of unmotivated land managers had such a large effect on the outcome was the result of a feedback loop. Unmotivated land managers are only a problem if they get infested in the first place. When there are more unmotivated managers in the landscape acting as a constant source, then it is more likely that other unmotivated managers will get infested and become sources themselves.