Apply for a research internship with the Conservation Decisions lab!
Sam Nicol and Iadine Chades are offering research internship projects for highly motivated students. These projects are suitable for students with a strong background in mathematics and proficient programming skills. In the past we have done some amazing work with students from French engineering schools. This year again, Simon worked on optimizing decisions to help the recovery of sea otters when an important parameter is unknown (ENSTA) and Martin (Ecole des Ponts) provided a novel method to help managers tackle the invasive Asian tiger mosquito in the Torres Strait Islands. Simon was selected for the best internship at ENSTA and Martin is applying for a PhD scholarship. It has been very rewarding to have them in the team and we really look forward to following their adventures.
See below a description of the broad context of our research. If you are an excellent student and would like to know more, send us your an email: firstname.lastname@example.org
Because resources to protect biodiversity are limited, conservation management must be cost-effective. We must achieve biodiversity conservation goals at a minimum cost. Being cost-effective means that our management decisions must anticipate what may happen in the future and maximise our chance of success over time. For example when conserving a population of an endangered species we must anticipate future changes in the abundance of the population. Population dynamics of endangered species evolve over time and are non-stationary: they are highly sensitive and can fluctuate in response to catastrophic events (drought, flood, fires) as well as human disturbances (habitat loss, degradation) and changes in climate. Management decisions might help recover our endangered species but their success is not guaranteed. To avoid species extinctions, we should account for all potential outcomes and costs in accordance with clearly specified conservation objectives before deciding on management actions. An example of a conservation objective is to maximize the chance of survival of an endangered species over the next 50 years using our limited budget. To provide informed guidance to managers, we must optimise our management decisions taking into account the uncertainty surrounding the response of the species to management actions as well as the economic costs and ecological benefits of the action over time.
Uncertainty is an inescapable aspect of natural resource management. While the uncertainty surrounding the population dynamics of species and the efficiency of management can be predicted using stochastic processes, finding the best management strategy over time requires a stochastic optimization procedure. Such optimization procedures are often referred to as stochastic dynamic programming (SDP) or Markov decision processes (MDP). Classic SDP techniques suffer from computational limitations that reduce their applicability to most ecological problems. Artificial intelligence and operations research have provided alternative solutions to tackle these impediments. Indeed optimization under uncertainty is one of the fastest growing areas of research where applications are mainly driven by mobile robotics or health systems. Ecological problems are similar in some modelling aspects, but they differ from other applications due to the specific constraints and solutions ecologists and decision-managers aim for. While an obscure optimal strategy will be suitable for a mobile robot on Mars, only meaningful rules of thumb are useful to ecologists and decision managers. Extracting these rules of thumb is a difficult exercise that requires interdisciplinary skills. First, a strong knowledge and understanding of the ecological system is required to frame the problem in such a way that the solutions will bring new insights to managing the system and can be generalized to similar problems. Once the problem is put into a decision-theory framework the optimization procedure should provide not only one solution but a set of equivalent solutions to choose from. The solutions should exhibit clear patterns that reflect the ecological problem, e.g. changes in management decisions should be fully explained by changes in the population dynamics of the species and the management efficiency. This is the aim of the applied research conducted by Dr Sam Nicol and Dr Iadine Chades at CSIRO.