One central question in biodiversity conservation is to identify the risk of species extinction. What is the probability of a population of a species to go extinct? How will species communities respond to ongoing global environmental change? Can we anticipate the risk of a community experiencing a strong shift?
The answer perhaps lies in probing the resilience of an ecological system by looking at its macroscopic responses to stress by using a set of generic resilience indicators (or early-warning signals). These indicators are based on dynamical systems theory, which states that when a system is close to a threshold it becomes slow in responding to disturbances. Resilience indicators can be directly quantified in the monitored time series of a system as increasing variance and autocorrelation with a set of out-of-the-shelf methods.
In particular, theoretical and empirical studies have shown that resilience indicators can be useful in anticipating abrupt ecological shifts in a variety of ecological situations. Drylands shifting to deserts at reduced levels of rain, lakes turning eutrophic due to increase in phosphorus nutrient levels, or fisheries populations crashing due to overexploitation, all can leave fingerprints of increasing variance and autocorrelation before the shift. These ecological shifts belong to a special class of responses called critical transitions. Critical transitions describe abrupt responses triggered by only small changes in external conditions, like the capsizing of a canoe caused by only a slight increase in tilting on one side. Once they occur, critical transitions are hard (if at all possible) to reverse.
Species communities can also experience critical transitions. Habitat destruction, extreme events, invasion of alien species, climatic changes could all cause abrupt changes in a community ranging from shifts in species dominance to species extinctions. For example, in food-webs (communities where species, like plants and herbivores, are connected through trophic interactions) the loss of a top predator could have a cascading effect leading to sharp changes in species dominance or even cascading extinctions. So how do we know if a community runs an increasing risk of a critical transition under some type of environmental stress?
The answer appears simple. We could monitor the populations of all species in a community and by using resilience indicators we could quantify temporal changes in stability to evaluate the risk of the community to experience an abrupt transition. But is that really possible? Communities are made up of only a handful up to hundreds of species. Continuous monitoring of a species in a community is a hard task, and even if we could continuously monitor a single species, it will be practically impossible to do to the same for all species in a community. But if — due to these constraints — we could monitor only a few species, how do we know that we measure the species that reflect the best the resilience of the total community?
To answer this question, we conducted a theoretical experiment by measuring resilience in simulated multispecies communities. Specifically, we investigated which properties defined the species that showed the strongest change in resilience indicators at increasing levels of stress. We did this by simulating the dynamics of communities of competing species for a common resource. Each community had a different architecture in the sense that the topology of the species network differed in its properties. We gradually stressed these communities until an abrupt shift in community composition took place. For each level of stress, we estimated changes in variance and autocorrelation using the biomass of each species as well as the biomass of the total community.
As expected, we confirmed that not all species in a community are equally sensitive in signaling the proximity to the community abrupt transition. Some species are better indicator-species than others. On the other hand, indicators based on combining all species biomasses perform better than species level resilience indicators. This result resonated earlier modeling work on mutualistic communities of plants and their insect or bird pollinators, which demonstrated that the strength of resilience indicators differed across species in a community prior to a critical transition.
Looking closer at the properties of the best and worst indicator species, there was no consistent pattern in their structural properties, like the number of species they interacted with or the architecture of the community they belonged to. Instead, it was the invading and collapsing species in a community that exhibited the strongest trends in resilience indicators prior to a transition. This implies that the species that are the ones to experience the strongest change are the ones that are driving the dynamics of the community and, as such, they could best reflect the upcoming change.
To a certain extent, this is logical to expect. Nonetheless, there were species that suffered big changes but did not reflect them in their resilience indicators. This was simply because their abrupt change was caused by species that were actually causing the change. Still, what makes species that “truly” drive the change and are the best indicator species is a mix of their structural and demographic properties that are difficult to a priori recognize, but it could be indirectly inferred by comparing the rate of change in resilience indicators among monitored species.
Understanding which species can be used as best-indicator species is a hard task. Nonetheless, being able to identify what species properties can make a species a “canary in a coal mine” could greatly advance and facilitate our current conservation programs contributing to the protection of fragile ecological communities under stress.
These findings are described in the article entitled Identifying best-indicator species for abrupt transitions in multispecies communities, recently published in the journal Ecological Indicators. This work was conducted by Vasilis Dakos from ETH Zurich and the Université de Montpellier.
- Dakos V (2018) Identifying best-indicator species for abrupt transitions in multispecies communities. Ecological Indicators 94, 494–502. doi.org/10.1016/j.ecolind.2017.10.024