Combining cost-effectiveness with complementarity
Evaluating the success of a conservation strategy is a crucial element of best practice management. Without it, managers can’t benefit from the experiences of others and scarce funds available could be wasted. And, if the strategy isn’t actually working, a lack of evaluation could lead to misguided policy directives and a loss of confidence of donors.
When it comes to evaluation, the challenge is always designing and implementing effective monitoring programs when funds are limited. Decisions need to be made about how, where, and what to monitor. To deal with this problem, many have suggested monitoring just one or a few indicator species rather than many species (see ‘Selecting good indicators’ in Decision Point #36). However, the vast literature on selecting indicators usually ignores one of the basic motivations for their use – provision of cost-effective information on whether an action is working. Decisions on what to monitor are routinely made ignoring the costs and benefits of alternative choices, and many organisations either end up monitoring everything (wasting lots of money that could be spent on management), or end up doing no monitoring at all (for fear of making the wrong choice).
The challenges of choosing indicators
Choosing which species to monitor is a big challenge. In part that’s because of uncertainty surrounding management outcomes. We often have very little idea of whether the action that’s been implemented will be effective. This should not stop us from acting or evaluating the impacts of our actions. But it does mean we need to ensure we account for this uncertainty in our decision-making.
There are two kinds of errors we can make when choosing indicators. The first is thinking an action is working when it isn’t. The second is believing an action is not working when it is. Each has consequences that may differ depending on the change we are trying to monitor and the number or characteristics of species being represented by the indicators. In a threatened species management context, thinking an action is working when it is not (ie, the species is still declining) could lead to the loss of a species because we fail to take further action. Alternatively, we might manage the system, fail to detect recovery of
threatened species because of insufficient monitoring (or monitoring the wrong species), and stop management prematurely.
Either way, the consequences of these errors can be catastrophic. Consequently, we need a way to select an indicator species that accounts for these uncertainties. Using decision science, we have devised a new way to do just this, a method that combines the uncertainties and costs. We demonstrated the value of our new method by applying it to a situation in the south-west of Australia, a region which has been declared a biodiversity hotspot due to its high plant endemism and high number of threats to declining biodiversity (See Decision Point #73).
One of the threats in this region is the introduced red fox preying on native mammals. Since the 1990s an extensive baiting program of foxes has been carried out by the Department of Environment and Conservation’s Western Shield program. The aim is to restore threatened mammal populations in WA’s reserve system. It has also recently been proposed by non-government organisations in private reserves to help protect fragmented remnant fauna populations. With millions of dollars
required to monitor and manage all fragments and all species, decisions need to be made about which species will tell us the most about the effectiveness of poison baiting for recovering declining populations of native fauna.
We selected fourteen mammal species from the south-west Australian landscape as possible indicators and we evaluated them for monitoring the effects of the fox baiting program. We had previously developed an approach for selecting indicator species that ranked each mammal on the basis of its cost-effectiveness for monitoring fox management (Tulloch et al. 2011), but a simple ranking approach does not account for complementarities between species.
“The principle of complementarity allows us to choose different species, each of which provides information on other species.”
In the case of indicator selection, the principle of complementarity allows us to choose different species, each of which provides information (eg, on behavioural ecology, habitat use, or responses to management) on other species, by measuring the extent to which one species contributes unrepresented values to an existing set of species. For instance, if two species provide the same information on responses to a given management action, do we need to monitor both of them, or would it be more sensible to try to find two species that respond in different ways? Because species often respond in different ways to a
given management action (or might not respond at all), it is wrong to assume that all will respond positively without evidence. To date, no indicator-selection frameworks have accounted for all the uncertainties in management outcomes as well as the potential for complementarity between species. We therefore decided a new framework was needed.
How decision science helps
Our new decision-science approach to cost-effective monitoring consisted of six steps: (1) define monitoring objectives and constraints; (2) list candidate indicators and calculate costs of monitoring each; (3) define data underlying species responses to management and determine the likelihood of detecting a trend when management is undertaken; (4) determine species surrogacy values (how representative one species is of another); (5) combine information on trend detection and surrogacy to calculate monitoring benefits (the amount of information each species will provide on another); and (6) solve optimisation problems.
We realised a simple benefit function was required that was easy to apply to any management problem, yet still accounted for uncertainty in the way it evaluated the monitoring power of an indicator. The benefit of monitoring a species was therefore the combination of the likelihood of successfully detecting a real change in the indicator, and the probability that this indicator successfully represented a set of target species. The benefits of monitoring every species for every other species can be represented in a network diagram that at first look appears complex and impossible to use for decision-making (Fig. 1).
Because this information is difficult to digest and use in its raw form, we set a range of expected responses and simplified the network to find the set of species that best represented that response (eg, Fig 1). By maximising the benefits of monitoring for a given expected response, we could then easily and transparently select the set of indicator species with the highest monitoring power for a given budget.
There is no single ‘best’ indicator species
By setting clear objectives for species recovery during invasive predator control, which included positive and negative target growth rates, we found that the likelihood of detecting a response to management changes depending on the desired direction and magnitude of the response. Different sets of species were therefore useful for telling us about different types of responses. We used published and unpublished time-series of responses to management to determine the benefits of
monitoring each species, but the quantitative benefit function that we developed is flexible for multiple scenarios and types of data.
Our new benefit function showed that the species currently monitored in our case study area were not always the most representative, cheapest, or most informative indicators of the responses of species to invasive-predator management. A number of species were either too variable in their expected response (eg, southern brown bandicoot) or had a high likelihood of showing a response in the absence of management (eg, western quoll). The two species selected most frequently for detecting an increase under fox management were the tammar and western brush wallabies. These species, although rare, are not listed as threatened but they have low variability and high growth rates when managed. The dibbler was also frequently selected as an indicator (Fig 2). This species is highly threatened and found in few locations, but its high certainty of response to management meant it was often selected over other species with higher surrogacy values.
Monitoring is a critical step that allows managers to learn from their experiences and to adapt management practices to maximize their effectiveness. Past research on indicator selection has ignored the varying costs of monitoring different species. It has also failed to account for uncertainties such as the likelihood of species responding to management, ability to detect real rather than spurious trends in populations, and how well one species represents another. If costs and species complementarity are not incorporated into the planning process, decisions could be costly and inefficient and uninformative
species might be monitored.
Many actions will not benefit all species. In our case study on invasive predator management, we found a likelihood of negative effects on some species. In these cases, we recommend a risk-averse strategy of selecting the set of species that maximises the expected benefit of detecting any change (negative or positive) that informs actions relevant to most species.
Our new framework incorporates all these components in a transparent benefit function and can be used as a model for decision makers to select a set of species for monitoring that provides the most reliable information on the responses of other species that cannot all be monitored due to funding constraints. This framework has the potential to enhance the utility and transparency of monitoring programs in the future.
More info: Ayesha Tulloch firstname.lastname@example.org
Tulloch AIT, Chadès I & HP Possingham (2013). Accounting for Complementarity to Maximize Monitoring Power for Species Management. Conservation Biology 27, 988-999.
Tulloch A, HP Possingham & K Wilson (2011). Wise selection of an indicator for monitoring the success of management actions. Biological Conservation 144, 141-154.