href="#u42bc1750-10ab-5a8a-a3d7-0328634fc8e9">Chapter 6, this volume) among a group of stakeholders or experts, may be useful for developing system health maps. PSM steps have been previously outlined (Sedlacko et al., 2014) and include defining the scope and boundaries of the system, system components, causal pathways, feedback loops, implications and knowledge gaps. The process may also be particularly relevant to development of system health maps because it allows for expression and inclusion of human values and a diversity of worldviews through development of mental models (Sedlacko et al., 2014). The National Cancer Institute’s (2007) report on tobacco control and the UK government’s obesity report (Butland et al., 2007) provide two detailed descriptions of systems map development and their use in directing needed interventions and policy changes for complex public health issues.
Once the system health map is developed, it will provide the structure for conducting health assessments and monitoring the effects of actions intended to improve system health. It is important to note that our understanding and, as a result, the map of the system will inevitably be imperfect and initially overly simplistic with missing components, unknown linkages and feedback loops, missing stakeholder groups, and misconceptions of system functioning. However, the system map can always be updated and improved as knowledge and understanding of the system grows.
System health metrics
Mapping the system at the scale of interest is the basis for development and selection of metrics. As each map is likely to be unique, a universal set of metrics is not possible. However, it will be important that metrics are based on the system map and correspond to the processes that are depicted. In other words, the metrics are an assessment of the system’s functioning as characterized by the degree to which each critical process is operating as expected or desired.
Many of the system’s processes can be measured in terms of rates (Fig. 3.3). For example, survival, immigration, emigration, reproduction, disease transmission and even stakeholder satisfaction are generally quantified in terms of their respective rates. Rates can be powerful for conducting assessments because there is often associated literature to help determine whether the measured rates are a positive or negative indicator of health. Rates can also be used as inputs into mathematical models to forecast the trajectory of the system. They allow direct comparisons between systems and can be used to measure changes over time. Similar to other studies of natural systems, the presence of epistemic and linguistic uncertainty warrants acknowledgement for system health assessments because factors such as population size are usually not precisely known and yet are needed as a denominator for many rates. Reviews of common types of uncertainty in studies of natural systems and methods to address them have previously been published (Regan et al., 2002; Milner-Gulland and Shea, 2017).
When system processes are difficult to measure directly, indices may be useful for indirectly measuring the process (Johnson, 2008). Indices can be quantitative or qualitative and should be used with caution because they rely on assumptions about how they relate to the process of interest. They may also be less sensitive than direct measurements or show some lag in responding to system dynamics. Despite these drawbacks, indices can be useful tools and, in some cases, may be the only means of assessing a particular process. An example of an application using indices as metrics of system health is the United States Department of Agriculture’s (USDA) tool for assessing the health of riparian systems (USDA, 2012), which uses a series of qualitative rankings associated with key processes. Questions in the tool include asking the user to rank the degree to which the riparian vegetation is composed of noxious weeds because their ‘presence or occurrence ... usually indicates a downward trend in ecological condition and riparian health’ (USDA, 2012). Regardless of whether the assessment tool is quantitative or qualitative, it will be important to continue to recognize that different stakeholders may have different thresholds for acceptable levels of change or system function.
Developing metrics to adequately assess how the system will respond to changes can be challenging, but it is critical that the ability of the system to adapt to change or the amount of change a system can absorb before it moves into a new state (i.e. resilience) be included in health assessments. Natural systems, regardless of ecological scale, are dynamic, and their ability to adapt to change is a key attribute of whether the system is fully functioning (Holling, 2001). Resilience of a system can be assessed by assessing its response to a perturbation (Holling, 1973). However, it is generally impractical or even unwise to experimentally perturb a system, so assessments require carefully choosing aspects of the system that provide evidence of how the system might respond to perturbations. This is where a system map becomes invaluable, because, if it has been constructed properly, it should highlight the processes that will drive change and associated system responses. For example, if we are assessing how a forest ecosystem will respond to fire disturbance, we might assess the quality of the seed bank, survey for the presence of invasive species, determine the degree to which species within the system are adapted to fire, look for evidence of past fires to understand the periodicity of fire in the system, quantify fuel loads, and so on. Using the information from various lines of evidence within the system, we can specify where in the adaptive cycle (Holling, 2001) the system currently lies and deduce how resilient the forest is likely to be to a new fire disturbance. We can also look for surrogate systems that may be similar to the system of interest but have gone through recent disturbances, and then draw conclusions from how the surrogate responded and apply them to how our system might be expected to behave or adapt under similar perturbation. Another useful set of tools for measuring resilience are mathematical models and computer simulation techniques that help forecast system responses to future change (Huyvaert et al., 2018). These models could be built from our system map, with system processes connected through specified mathematical relationships, and parameterized based on the values for the various metrics we described earlier. The simulation or model is then run and resulting trajectories for the system capture not only the current state of the system but also future states if processes/metrics remain static. The adaptive ability of the system can be assessed by varying system processes based on their likely response to some change and quantifying how the new forecasts of system states have been altered compared with forecasts when processes remained static. Regardless of the chosen tool, assessing the system’s adaptability cannot be ignored because it will have direct implications for system functioning and its associated health.
Once each of the system processes have been measured either qualitatively or quantitatively, processes that are not meeting stakeholder expectations or may be negatively impacting the health or function of the system will need to be highlighted. For these processes, a closer examination of why the process is not functioning to its expected potential is required. This entails examining the system map and identifying the biotic and abiotic components tied to these processes and determining where there are deficiencies. For example, if juvenile recruitment of a population is too low, resulting in forecasts of future population declines, a search for deficiencies in critical components will need to be made (e.g. a lack of thermal cover leading to nest failures). The identified deficiencies can become the target of future management actions to improve health and could drive the final recommendations arising from the health assessment.
A practical
