There is a saying that those who forget history are destined to repeat it. Thus, we often look to the past to guide our actions in the present. However, this is not always as straight forward as it sounds. My colleagues and I wrote two papers based on a historical dataset of exotic grass introductions into Australia. In one we untangled the effect that various economic, social and policy changes had on exotic plant introductions (van Klinken et al. 2014), and in the other, we used the benefit of hindsight to test which types of grasses went on to have a higher impact (van Klinken et al. 2013). Can these findings help make decisions about managing invasive species in the present? In my opinion, they only provide limited assistance with specific management decisions.
Taking general lessons from history is relatively straight forward. Our papers highlighted broad historical patterns with some relevance for the present. We found that most species which were introduced did not become particularly problematic, and those that did, tended to be semi-aquatic and wide-spread. We also found that since the Australian government started restricting plant importations in the post war period, the number of new species being introduced has reduced dramatically. As a result, all of the species in our dataset that have a substantial negative impact have been present in Australia for at least several decades. This suggests three sensible, if obvious, lessons. Firstly, it would be inadvisable to import new species with the same enthusiasm as during the first half of the 20th century. Secondly, it is likely that the introduction of new, damaging species will be rare over the short to medium term future and most problems will be caused by species already in established Australia. Thirdly, the way in which we weight the risk of misclassifying a high impact species as a non-damaging species changes how we should predict which species will have a high impact.
Our papers demonstrate how historical data can be used to make predictions. As predictive tools they are reasonable. In the case of van Klinken et al. (2013) the best model is more than twice as good as random guessing (see Table 3, van Klinken et al. 2013). However, predictions alone are not enough. There is an entire branch of mathematics called decision theory, which establishes how to make a good decision. Under a decision theory framework there are seven steps to making a management decision, and only one of them provides prediction. Yokomizo et al. 2014 outlines these seven steps:
1. Specify the management objectives
2. List the possible management actions
3. Specify the variables that describe the state of the system being managed
4. Develop a model of the system that predicts how each management action will affect the system (this is the step where predictive models are needed).
5. Specify the constraints on actions (e.g. we don’t have enough resources to do all actions) and states
6. Specify the magnitude of uncertainty (predictive models can help here as well)
7. Find good solutions to the problem.
This list illustrates that good decisions are specific to the goals and contexts in which they are developed. A context in which our predictive models may be useful is bio-security. We often have to decide if the benefits of introducing a new species will outweigh the danger that the species will become invasive and cause widespread damage. In order to calculate the chance that any given species will go on to become problematic, we need a predictive model. However, as seen in the list above there are six other factors that need to be addressed to make a good decision. The other steps, such as establishing goals, defining actions and setting constraints, cannot be solely answered with science. These steps require the input of managers, policy makers and society as a whole. It is these groups that ultimately decide what levels of resources are available, what actions are possible and what is socially acceptable.