2.5.1 Early warning signals in social-ecological systems: The challenge

Substantial research has demonstrated the potential for early detection and anticipation of tipping points in ecological systems and Earth system processes (see Chapter 1.6). The basic principle is that additional stress can transition an environmental system from one ‘potential well’, such as a tropical forest, to another – for example, a dry savannah. The social-ecological systems impacted by Earth system tipping points can themselves behave in a similar manner, whereby continued environmental stress can lead to (practically) irreversible changes in socioeconomic conditions (see also Chapter 1.6 for an overview of early warning signals for Earth system tipping points). Pastoralist systems serve as an illustration of how such transitions can occur: the addition of livestock in a grazing land might degrade pasture and cause accelerated soil erosion, permanently transforming the landscape from a fertile pasture to a semi-arid shrubland or even a desert, and rendering traditional pastoral livelihoods unfeasible (Feng et al., 2021; Ibáñez et al., 2007). Chapter 1.6 presents analogous examples in environmental systems.

However, social-ecological systems are highly complex and do not always exhibit traditional bifurcation and early warning signals, which may provide misleading results. As such, before designing early warning systems it is important to understand the nature of the hazard, the vulnerabilities being driven from both social and biophysical drivers, exposure to risks, and whether the system can exhibit signs of bifurcation. For instance, in the case of social media, high autocorrelation of tweets might be interpreted as an early warning signal of a tipping point, when in reality the autocorrelation trend can be explained by knowledge that a specific event or holiday is approaching (Bentley et al., 2014; Kuehn et al., 2014).

While there is potential to borrow and adapt elements from traditional tipping point theory (which focuses on ecological applications), there are a number of considerations in social systems. First, social systems have features that cannot be compared to those of environmental systems, limiting their predictability (Milkoreit et al., 2018). For instance, even when comparing two communities in the same country there will be differences in power structures, access to information, economic equality, engagement in decision-making processes, knowledge, and capacity to adapt to changes, all of which can affect the manifestation of a tipping point in a social context. Second, continuous data in social systems are not always available. Often social elements are rather abstract – even if an adequate indicator or proxy is identified, it may not be feasible to collect data over time to enable detection of tipping points. Moreover, social science methods such as ethnographies, interviews, surveys, and focus group discussions are expensive and time-consuming; as such, they tend to be ad hoc and of insufficient temporal resolution to identify critical transitions (cf. Shipman, 2014).

(Milkoreit et al., 2018) illustrate the complexity associated with detection of tipping points in social-ecological systems using resource extraction as an illustration. In a fishing context, an ‘ecological’ regime shift might be the collapse of fisheries as measured by fish stock, the health of the local coral reefs, or even water quality. A social tipping point might be a collective decision to engage in alternative livelihoods and reduce (or altogether cease) fishing. In turn, the local identity as a fishing community might change to an entirely different social state. The first tipping point (the decision to engage in non-fishing activities) could be measured through various surveys and livelihood assessments, while the second, more abstract indicator (community self-identification) would require qualitative methods. In both cases, it is unlikely that regular data would exist to determine when exactly the transition has occurred. Research has shown the potential to quantify tipping points in emotional states through temporal autocorrelation, variance and correlation of self-recorded emotions (van Leemput et al., 2014). So, elements of tipping point theory may be applied to more abstract concepts that are pertinent for social science application, depending on data availability. For example, Koschate-Reis et al., (2019) have shown that tipping point theory can be applied to automatically detect feminist/parent identities from textual data.

The data challenge is significant – however, recent research has shown the potential to use polling data (Winkelmann et al., 2022), online surveys (Ehret et al., 2022) and Earth observation (Swingedouw et al., 2020; Krishnamurthy et al., 2022) to provide early warning signals of tipping in coupled social-environmental systems. For data to be maximally useful, they need to be available at an appropriate frequency to enable analysis of system dynamics. For example, in a food security application of tipping point theory, (Krishnamurthy et al., 2022) determined that data from the Soil Moisture Active Passive (SMAP) Earth satellite mission – which are available every 3.5 days – were the most appropriate for detecting an impending food crisis. Datasets available at coarser temporal resolutions (including other soil moisture products and vegetation health indices) were less accurate for early warning signals, though future observations at higher spatiotemporal resolutions and accuracies may improve results even further.

Another significant challenge is interpretation of false positives. Predictions of catastrophic change – such as Ehrlich’s (1968) predictions of famine due to excess population, or peak oil in the 1990s and 2000s (Bardi, 2019) – have failed to materialise, creating a public sense of mistrust. While early warning systems are extremely useful to anticipate (and avert) the worst effects of climate change, the history of high-profile false positives has created an easy target for critics seeking to belittle social risks associated with climate change (and other environmental crises).