Two useful tools for resilience-building in social-ecological systems, are structured scenarios and active adaptive management (Olsson et al., 2006). These consider the alteration of the magnitude, frequency, and duration of disturbance regimes to which the system is adapted (Radhakrishnan et al., 2017). Due to a lack of data on the local scale we include a model framework that relies on regional projections. One of the resilience frameworks originates from risk and uncertainty assessment under climate change.

This method enhances reproducibility and is applicable to water, land, and ecosystem management adaptation (Werners et al., 2015). With the Adaptation Tipping Points (ATP) framework policy and management actions are determined when “the magnitude of change (e.g., climate change) is such that the current management strategies are deemed to be inadequate to meet the social-ecological objectives”(Kwadijk et al., 2010).

The application of this framework requires information about water management and land-use planning strategies (Rijke et al., 2013; Werners et al., 2013a). The advantages of applying this method, include stakeholder engagement and the definition of objectives and aims in management of the SES. Challenges remain to overcome when applying the ATP method to ecosystems. These include a lack of data availability and unclear or undefined ecological thresholds (Bölscher et al., 2013; Werners et al., 2013a,b; Nanda et al., 2018a). Overcoming these challenges will help to shape adaptation strategies that are essential to transit to resilience climate change strategies for ecosystems (Haasnoot et al., 2012; Ahmed et al., 2018). The five-step ATP methodology includes: (i) the determination of climate change effects on the system; followed by (ii) identifying key objectives and thresholds; (iii) the determination when standards were compromised in the past; (iv) analyzing when standards will be compromised in the future; and (v) to repeat step 1–4 for alternative strategies (Kwadijk et al., 2010).

The original methodology was modified in step 3. We identified eight ATPs for different social-ecological objectives and thresholds with the assessment of historical hydrological time series in a previous study (Nanda et al., 2018a). Step 3 was extended to interpret ATPs in conjunction with the hydrological response and variation; temporal scale ecosystem responses; and recovery rate and alternative stable state of ecological processes (Figure 1). Steps 4A till 4C were also conducted in a previous study (Nanda et al., 2018b) and focused on governance of the SES which described the mismatches between the ecosystem organization, jurisdiction, and institutions. Based on the previous results we focus here on a repeat of steps 4A and 4B in which we focus on adaptation measures that go beyond the eight ecological objectives (Supplementary Tables 1–4).

Interviews from previous studies (Nanda et al., 2018a,b) were used to add ecosystem processes/patterns caused by alterations of the hydrological system, mainly due to decreased rainfall. From observed seasonal surface- and groundwater levels of Lake Forestdale we created a conceptual eco-hydrological model that highlights the seasonal processes of the wetland. To illustrate the ecological implications of lower surface water levels we collected information about management implications (from stakeholders) of the deep water and shallow water zones and the wetland buffer zone.

As wetlands of the SCP are often steppingstones in linear corridors in the landscape, adaptive management focuses on the integration of water resources and ecosystem management. Therefore, the wetland buffer zone guidelines and ecological water requirements are explained with the available literature that draws on the theories of resilience and ecology. The objective to maintain a wetland buffer zone defined in scientific literature, depends on radial areas measured from the furthermost extent of wetland vegetation and depending on the width of the zone to protect the integrity of the wetland (Castelle et al., 1994). We rated the buffer zone according to main landscape components (Ignatieva et al., 2010): (1) patch quality (habitat value for species), (2) boundary effects [the surroundings of a patch that influence movement of species and protects from disturbance (Weston et al., 2009)], (3) patch context (the spatial location of a patch), and (4) connectivity (movements of species between patches).

Step 4C incorporated the use of eco-hydrological conceptual models of the case study area in interview questions. These were aimed to complement the eco-hydrological conceptual model and inform whether adaptation measures were beneficial for one SES objective and simultaneously negative for another (maladaptation).

The SCP wetlands in southwest Western Australia provide habitat for endemic flora and fauna and they are located within one of the international biodiversity hotspots (Myers et al., 2000). Forrestdale Lake is a Ramsar listed wetland with extensive fringing sedgeland which is typical of the Swan Coastal Plain (Figure 2). In a regional context, the lake is a major breeding site, migration stopover and semi-permanent drought refuge area for waterbirds and migratory birds. The lake contains open water and is fringed by rushes and a continuous belt of bulrushes of rare native Typha domingensis and introduced/invasive Typha orientalis, behind which are belts of trees tolerant of water-logging (CCWA, 2005). The higher ground around the lake supports open woodland while the aquatic zone consists of macrophytes. The margins of the lake support many terrestrial birds and other vertebrate species, such as the quenda (Isoodon obesulus fusciventer) and invasive European red fox foxes (Vulpes vulpes).

The sediments of the SCP are made of wind and riverine depositions. These sediments form four main geomorphic units with: distinctive geology, topography, drainage patterns, soil characteristics, and vegetation (Semeniuk, 1987). Wetlands in the SCP lie in depressions between dunes which are located where a shallow water table permanently or seasonally intersects the land surface. The depressions appear to act as flow-through lakes, which capture groundwater on the eastern side (Darling Scarp) and discharge water westwards toward the Indian Ocean (Townley et al., 1993).

The geomorphic elements of the Southern River Catchment are typical of the SCP located in the Perth Basin. On the eastern edge of the Perth Basin are the Bassendean Dunes which comprise sandy soils. Lakebed sediments comprise sand to sandy organic mud overlying soft marly limestone and clayey sand (McArthur et al., 1980; Davidson, 1995). The Jandakot Groundwater Mound is situated in the managed Perth Groundwater Area, which is a region of elevated groundwater table beneath the SCP. Groundwater discharges from the mound into low lying depressions that support groundwater dependent vegetation and extensive wetland systems. Wetlands are surface expressions of groundwater where a shallow water table intersects with the ground surface (Semeniuk, 1987; Hill, 1996).

Wetlands of the SCP are classified by water periodicity (permanently or seasonally inundated/waterlogged) and by landform (basin, flat, channel, slope, or highland) with damplands, sumplands, and palusplains the most frequent occurring type of wetlands (Semeniuk, 1987; Semeniuk and Semeniuk, 1995; Hill et al., 1996). Seasonally inundated refers to wetlands depending on rainwater (surface water), while seasonally waterlogged wetlands depend on groundwater which have soils that are saturated throughout summer). Further differentiation between wetlands of the same classification is possible using wetland descriptors such as water quality, size, shape, and vegetation. The wetland types found in the SCP vary in size, shape, hydrology, stratigraphy, and vegetation. The most frequently used classification is shown in Table 1. Wetlands can move through different hydroperiod states as a result from declining rainfall and increases or decrease of the groundwater table. Forrestdale Lake is a groundwater lake and surface run-off had little effect on its depth when it was in an undisturbed condition. Alternative system states have not been extensively studied and are often limited to the hydroperiod of a wetland (Davis and Brock, 2008; Davis et al., 2015).

The wetlands of the SCP are likely to experience increasing stress, as the population is expected to double by 2050 to 4 million inhabitants (Australian Bureau of Statistics, 2013a,b). With a growing population demand will increase for (ground)water and land for development, posing risks to hydrological alteration that support the critical ecological processes. Climate change studies predict decreasing rainfall for southwest Western Australia adding to an increase of hydrological stress (Smith and Power, 2014). Wetland management needs to respond to a drying climate so that the integrity of the wetland's ecosystem functions is maintained. The responsibility for managing the wetland is shared among different actors (e.g., governments or communities).

The ideal situation for adaptive ecosystem management, is to take timely decisions that maintain systems in desirable stable states that provide the necessary ecosystem services such as biodiversity and to keep this system for the future (Folke et al., 2005; Schultz et al., 2015). This is consistent to systems behavior, which after perturbation, can shift systems in alternative and often undesirable states with few ecosystem services (Scheffer et al., 2001; Folke et al., 2004; Carpenter et al., 2015). The uncertainty about systems behavior and perturbations are major challenges for ecosystem management (Berkes, 2007). Different system states are difficult to determine and often are evaluated from limited data and qualitative assessments to distinguish important ecosystem processes. Once ecosystems have changed; it is often challenging to revert, restore, or remediate them (Folke et al., 2004).

The behavior of an ecosystem depends on many feedback processes that interact with the social system (Holling, 2001). As studies often focus on single drivers of system change, such as climate change, other drivers may be overlooked (Lavery and Donovan, 2005; Kwadijk et al., 2010; Bölscher et al., 2013). Ecosystem resilience is addressed by known system characteristics but becomes very complex when more drivers for change are identified. The interpretation of results requires careful consideration of other external drivers that are responsible for ecosystem performance and possible positive or negative responses (trade-off) within a SES (Cohen-Shacham et al., 2011; van Oudenhoven, 2011; van Slobbe et al., 2016).

Internal drivers could also be important to biophysical processes, such as diseases known to decrease plant resilience (internal) and future land-use change (external). Despite previous research which has pointed to climate change as the current main external driver for change, land use changes need to be considered in the context of adaptation strategies (Seitz et al., 2011; Westbrook and Noble, 2013). Land-use changes (e.g., urbanization and stormwater collection) could be a source of water and thus could support the ecosystem processes, although, under the event of extreme drying these adaptation measures may not be sufficient (Barron and Barr, 2009; Barron et al., 2013).

Involvement of all stakeholders of the water management system is crucial to deliver adaptive strategies and to overcome barriers by science informed planning (MacKay, 2006). To support ecosystem functions patches or habitats must be large enough to support viable populations of endangered species and to support complicated food webs. Planning (spatial and temporal) of these wetland ecosystem services must be coordinated by different jurisdictions under different institutions (Noble et al., 2011; Schleupner and Schneider, 2013).

Resilience of ecosystems is problematic to assess when past or future conditions cannot be compared to baseline ecosystem states (Bloemen et al., 2017). This follows from the definition of ecological resilience, which states that ecosystems are not in an equilibrium, but rather move from one to another stable state (Folke et al., 2004). This is contrary to management, that follows a one-direction and incremental approach (Olsson et al., 2006). However, this does not take into account that individual processes within a system can increase resilience beyond that single ecosystem (Reyers et al., 2018). For example, migratory bird species may find suitable habitat in other wetlands if the original wetland lost the ability to retain them. From a management perspective, this does not immediately lead to a response for restoring this ecosystem service, since sociological or economic considerations are also involved in the process of decision-making (Schultz et al., 2015). Individual processes need to be assessed regarding the likelihood that they can be sustained under extreme conditions (Hughes et al., 2017), so that management authorities can better prioritize their resources (Ostrom, 2009) and select ecological objectives that they are willing to support.

The principle of management for resilience depends on spatial and temporal scales, both for institutional and ecosystem processes (Young, 2002; Huitema et al., 2009). Especially, when these two processes are not similarly organized, assessments for ecosystem resilience must consider multiple spatial scales (Dewulf et al., 2015). Since the adaptive capacity of an ecosystem also depends on the decision-making process; institutional organization over spatial scales; and levels of governance must also be analyzed (Brooks et al., 2005; Pahl-Wostl, 2009). Quick wins and long-term solutions emerge from multi-scale analyses while implementation of adaptation measures on local level without the need for higher level interventions requires consensus (Fünfgeld, 2015; Moloney and Fünfgeld, 2015). Higher level legislative changes require more negotiation at institutional levels and may take longer for implementation. Future research needs to consider the decision pathways for different organizational structures. These may be different and require other institutions or jurisdictions to implement strategic decisions to create a decision space (Carpenter et al., 2015; Hermans et al., 2017).

Resilience and adaptive capacity provide information about the vulnerability of SES and closely link social and ecological interactions (Smit and Wandel, 2006; Berkes, 2007); even if clear feedbacks are not known. Feedbacks with actors provide implementation of adaptation measures on the ground which also need to distinguish slow and fast responses to changing systems when thresholds are exceeded (Walker et al., 2012). Even though thresholds are not “fixed” for biophysical processes; exceedance of thresholds can be interpreted with the available ecological data and ATPs to assess individual management objectives [Examples for salmon re-introduction (Hanna, 2008; Bölscher et al., 2013; van Slobbe et al., 2016)]. Future scenarios could inform management timely to prevent exceedance of individual objectives while collecting data and building knowledge with stakeholders is critical to identify information gaps of ecosystem functions and feedbacks.

Current thresholds and objectives give an indication of the ecological success of the ecosystem; however, the advanced conceptual understanding of the ecosystem was more effective to guide the process on effective (structural and non-structural) adaptation measures. Adaptation is managed through cross-scale interactions to avoid a decrease of resilience (Holling, 1996, 2001). In addition, actors in our case study could (i) move thresholds closer to the current state of the system by altering latitude (water suppletion) and (ii) make the threshold easier to reach by lowering these (Walker et al., 2004). Increased system understanding could be helpful to identify the impact of climate change on eco-hydrological processes, but there is still a challenge to identify when other parameters (e.g., water quality) become more relevant when a system may shift to alternate states. Social-economic concerns were not considered in our overview of responses to change, but research shows this is an important link (Biggs et al., 2012; Reyers et al., 2018). Also, we did not consider environmental and citizen groups that may have limited influence on the socio-economic considerations of government actors. Citizen groups often depend on the funding from government for ecosystem-based measures (e.g., removing weeds, revegetation).

Stakeholders represent multiple jurisdictions in the case study area and therefore are not limited to their own jurisdictions. Cross-scale interactions could enhance the connectivity of multi-level SESs (Brondizio et al., 2009; Biggs et al., 2012; Reyers et al., 2018). Cross-jurisdictional collaboration often lead to ecosystem-based considerations aimed to match with ecosystem organization (Young, 2006). Workshops can encourage discussions between government representatives, researchers, and environmental and citizen groups about (un)acceptable change and definition of critical indicator values (Ferguson et al., 2013; Chaffin et al., 2016). Projections in different time periods in the future, enables to provide the level of uncertainty for existing policy and to support new adaptation responses.

Without an integrated and coupled SES model, future decisions and ecosystem responses cannot be easily predicted. Often, much data is required, whereas a coupling between the social and ecological domains can be realized by engaging stakeholders. Resilience approaches deal with uncertainty and data gaps due to simplification of understanding of SES functions to build knowledge for conceptual and hydrological models (Sivapalan et al., 2012). The intensification of policy oriented and bottom-up processes could further increase cross-jurisdictional collaboration and further support the science-policy interface for adaptive ecosystem management.

Natural climate variability, patterns of land-use change, patterns of water regulation, structural design of infrastructure, and distal societal drivers, are just some of many elaborated processes that influence SESs. With drivers (e.g., climate) increasingly leading to changes of complex ecosystem processes, there is a risk that ecosystem management becomes increasingly misaligned with legal, social, ecological, and economic considerations. In the past-anthropogenic factors have already caused the complete collapse of ecosystems (Jackson et al., 2001). This makes the political and public support for ecosystem protection important, to prevent them from becoming lost entirely.