A consistent vegetation classification for wetland conservation and management in the Murray–Darling Basin, Australia

Water reform, wetlands and the Murray–Darling Basin Plan

The Basin in south-eastern Australia contains more than 30,000 wetlands, of which 16 are Ramsar-listed sites of international importance and over 200 are considered of national significance (Department of Sustainability, Environment, Water, Population and Communities 2011). Low-gradient, meandering rivers traverse extensive floodplains, with terminal wetlands at river confluences. Many wetlands are in semiarid and arid regions, subject to highly variable flow regimes and fluctuate between extended dry phases and short wet ones (Colloff and Baldwin 2010). These wetlands contain unique vegetation communities adapted to high evaporation and salinity and low water availability (Roberts and Marston 2011; Roberts et al. 2016) that are important refugia for biodiversity during prolonged droughts.

Basin wetlands are considered degraded and in most catchments they have been ranked as in poor or very poor ecological condition (Davies et al. 2010, 2012), owing to combined effects of river regulation, water diversions for irrigation, expansion of irrigated broadacre cropping, climate change and prolonged drought (Kingsford 2000; van Dijk et al. 2013; Colloff et al. 2015). The poor condition of wetlands prompted a major program of water policy reform, culminating in the Commonwealth Water Act 2007 and its statutory instrument, the Basin Plan (2012), with the objective of returning water that had been over-allocated to irrigators back to the environment.

The Water Act is underpinned by the principles of ecologically sustainable development and gains its constitutional legitimacy largely from the enactment of international environmental treaties, which require the conservation and sustainable use of wetlands. But commitments to the Ramsar Convention and CBD have been marginalised in the implementation of the Basin Plan (Kirsch et al. 2022; Bender et al. 2023). Ultimately, there is not enough water to meet irrigation and environmental requirements and what is available remains highly contested. Delivery of managed environmental flows has fallen short of what is needed to conserve wetlands and their biodiversity (Chen et al. 2021; Ryan et al. 2021).

Improving biodiversity conservation requires a robust, scalable, predictive, scientific framework on ecosystem responses to environmental change and management interventions (Keith et al. 2022). Such a framework is lacking for wetlands of the Murray–Darling Basin and has led to ineffective decision-making for determining priority conservation areas and how expensive, scarce and highly contested environmental water can be used to achieve the CARE principles of connectivity, adequacy, representativeness and efficiency. This knowledge gap in decision-making highlights the need for a practical, robust wetland classification that defines wetlands in terms of their ecological structure, processes and character to inform improved conservation policy and management. Our concerns over the suitability of the interim ANAE classification for the Basin (Brooks et al. 2014) to achieve this objective were a major motivation for the present research.

Our aim herein is to assess the suitability of existing wetland and vegetation classifications for the definition, delineation and inventory of Basin wetlands. This assessment includes (1) identification of the strengths and weaknesses of current systems by using the principles of adequacy of definition, consistency, information quality and reproducibility, (2) developing a standardised, reproducible classification for mapping and identification of representative wetland vegetation types, and (3) examining applications of this classification in wetland conservation decision-making, including management of environmental watering requirements, to align wetland policy and management with national and international conservation commitments, particularly the ‘30 by 30’ Kunming–Montreal conservation targets.

Principles used to assess wetland classifications

We used the following principles to assess wetland and vegetation classifications used in the Basin: adequacy of definition, consistency, information quality and reproducibility.

Assessment of wetland and vegetation classifications

The following six main classifications are used for wetlands within the Basin: the National Vegetation Information System (NVIS) (National Land and Water Resources Audit 2001; Department of Climate Change, Energy, the Environment & Water 2024a); ANAE classification (Aquatic Ecosystems Task Group 2012), including the interim Basin classification of Brooks et al. (2014), and separate systems for Queensland (Neldner et al. 2023), New South Wales (Keith 2004; Benson 2006), the Australian Capital Territory (ACT; Armstrong et al. 2013; Baines et al. 2013) and Victoria (Department of Environment, Land, Water and Planning 2016). South Australia has a regional classification for Murray River wetlands (Jones and Miles 2009), but has otherwise adopted the NVIS (Department for Environment and Heritage 2006). The ACT government uses the ANAE for wetlands (Environment, Heritage and Parks 2024) and the New South Wales classification covers native vegetation communities of the ACT.

We undertook an analysis of technical reports, manuals, websites and other publications for NVIS, the interim Basin-scale ANAE and the classifications for Queensland, Victoria and New South Wales. We collated data on their attributes and design principles, including the typological basis of their units, i.e. whether they were based on vegetation communities or also included functional attributes and hydro-geomorphological features as separate entities, as well as the structure or hierarchy of their units and any associated documentation on application and use of the classification. We then assessed each classification on the basis of the four assessment principles listed above. Finally, we identified the most suitable classification and applied it at Basin scale to map distribution and extent of wetland types.

Basin-scale integration and mapping

We selected the New South Wales classification at the scale of vegetation classes (Keith 2004; Benson 2006) as the most suitable, as detailed below in the results section. We then integrated the units of the other State-based classifications with those for New South Wales to produce a Basin-scale classification and map as follows: for Queensland, based on regional ecosystems (REs) and broad vegetation groups (BVGs) (Department of Environment, Science and Innovation 2024); for New South Wales, based on vegetation formations, classes and plant community types (PCTs) (New South Wales Environment and Heritage 2022; New South Wales Government 2024); for Victoria, based on ecological vegetation classes (EVCs) (Department of Energy, Environment and Climate Action 2024a, 2024b) and for South Australia, based on native vegetation floristic areas (Department of Environment and Water 2023).

To re-classify wetland vegetation into standardised, Basin-wide vegetation classes, we used the fine-scale vegetation units from the State-based classifications (REs for Queensland, EVCs for Victoria and Vegetation Code for South Australia). We assigned each wetland vegetation unit to the most appropriate vegetation class in the New South Wales classification, on the basis of similarities of floristic composition, community descriptions, distribution according to hydro-geomorphological features and other habitat attributes, and assigned them to an attribute table for each State (Supplementary Tables S1–S4). We then filtered these to select only wetland vegetation classes based on the New South Wales classification.

We used ArcGIS Pro (ver. 3.0.3, Esri, Redlands, CA, USA) to map wetland vegetation classes, using vector datasets, which delineate the vegetation communities as polygons. We projected vegetation datasets onto the co-ordinate system Geocentric Datum 2020/MGA 2020 Zone 55, by using the ‘project’ tool to maintain consistency. We clipped the maps to the Murray–Darling Basin boundary (Murray–Darling Basin Authority 2024a). Map units and their vegetation classes were compiled into.csv files for joining with the original dataset.

Some polygon data from Queensland and South Australia contained multiple vegetation units. In such cases, we retained polygons in which the wetland vegetation unit accounted for ≥70% of the area. Mixed polygons containing wetland vegetation units that accounted for 50–65% of their areas were included if the primary vegetation units belonged to wetland types. To address the absence of estuarine wetlands in the New South Wales classification, for South Australia, we merged polygons representing the Coorong, Lake Alexandrina and Lake Albert (Department of Climate Change, Energy, the Environment & Water 2021) map as one vegetation class.

We incorporated a layer of protected area extent (2022) from the Collaborative Australian Protected Areas Database, which includes public, private, Indigenous and jointly managed protected areas (Department of Climate Change, Energy, the Environment & Water 2023a). We used the extract-by-mask function in ArcGIS to delineate where protected areas overlapped with wetlands, to assess distribution and extent of wetlands in protected areas. Spatial distribution and extent of intensive agricultural activities were also mapped using data from the Australian Land Use and Management Classification (ver. 8; Australian Bureau of Agricultural and Resource Economics and Sciences 2018).

To compare the extent of wetland vegetation classes with the extent of the floodplain which receives environmental flows, we used a modified version of the layer entitled ‘Flow-MER MDB managed floodplain’ (Commonwealth Environmental Water Office 2022), which includes areas that received Commonwealth environmental water (2014–21). This layer is a modified version of the ‘managed floodplain with constraints relaxed’ layer (Murray–Darling Basin Authority 2023a), which includes actively managed areas that receive environmental water from headwater dams or environmental works, and ‘passively managed areas’ that receive environmental water by implementation of flow rules and can be flooded with high flows if key constraints are overcome or by natural events. We modified the Flow-MER managed floodplain layer to better reflect the area of what we refer to as the ‘actively managed floodplain’. We removed areas in the unregulated Paroo and largely unregulated Warrego catchments because floods along these ephemeral rivers depend predominantly on high natural flow events (Murray–Darling Basin Authority 2023b, 2023c), which cannot be considered as ‘managed environmental flows’. It is important to note that the boundary of the actively managed floodplain (Fig. 3f) is not fixed. It is subject to assumptions regarding actions that have yet to occur. These include when, and to what extent, constraints on environmental flows will be relaxed (Murray–Darling Basin Authority 2024b) and the effect of environmental flow rules in Water Resource Plans, which have yet to be implemented.

After re-assembling mapping units into the vegetation classes, we processed data using the geometry attributes function in ArcGIS to measure the extent of each vegetation class.

The NVIS classification

The National Vegetation Information System (NVIS) defines 32 major vegetation groups (MVGs) and 98 major vegetation subgroups (MVSs) (National Land and Water Resources Audit 2001; Department of Climate Change, Energy, the Environment & Water 2024a). However, individual MVSs are not fully nested within MVGs and are represented in more than one MVG in many cases (Keith and Pellow 2015). Wetlands correspond poorly with the structurally defined groups within NVIS. For example, Eucalypt Woodland (MVG 5), the most widespread MGV in the Basin, covers both wetlands and terrestrial environments, uplands and floodplains and includes a series of what would otherwise be considered quite distinctive eucalypt woodland communities. The four MVSs that correspond with MVG 5 (MVS 8, 9, 10, 65) are defined by broad categories of understorey type (tussock or hummock grass-, shrubby- or chenopod-dominated) that occur both in terrestrial habitats and floodplain wetlands. Accordingly, NVIS does not satisfy the criteria of adequacy of definition, consistency of application and reproducibility for wetland types. It fails the basic requirement of Pressey and Adam (1995) for lines on maps that clearly distinguish wetlands from other parts of the landscape.

The interim Basin-scale ANAE classification

The ANAE classification is a hierarchical, rules-based system with the following three scales: Level 1: regional; Level 2: landscape; and Level 3: aquatic system classes (Aquatic Ecosystems Task Group 2012). The interim Basin-scale ANAE classification (Brooks et al. 2014) is intended to be used to guide and assess environmental watering decisions on the basis of a measure referred to as ‘ecosystem diversity’ (Brooks 2022).

At the scale of aquatic system classes, the Basin ANAE classification uses the same categories as the US Forest and Wildlife Services classification (Cowardin et al. 1979). However, this classification was designed for northern hemisphere post-glacial landscapes with a predominantly temperate, wet-to-mesic climate. Water regimes and hydroperiod, geomorphological features, topography, climate and vegetation communities are completely different from those of Australia in general and the Basin in particular. The categories adopted from the US system (palustrine, riverine and floodplain) are less applicable in the Australian context because wetlands features in the Basin do not fit neatly into those categories. For example, on low-gradient floodplains (Kingsford et al. 2004), there is little or no differentiation between a lake bed (lacustrine), its swampy lakeshore margin (palustrine), adjacent floodplain and the riverine system (Semeniuk and Semeniuk 1997).

In the Basin, with very high inter-annual variability of rainfall and flows (Puckridge et al. 1998), the use of hydroperiod to define wetland types, i.e. permanent, semi-permanent, temporary, intermittent, as used in the ANAE classification, is rendered imprecise and inconsistent, especially under hydrological alterations as a result of climate change (Leblanc et al. 2012). Use of these terms ignores the marked variability of flows in arid and semi-arid regions (Roshier et al. 2001). Few wetlands in the Basin could be considered ‘permanent’ or even ’semi-permanent’ and have not been subject to either prolonged drying or considerable variation in flow regimes in recent decades. This situation contrasts markedly with North America, with an abundance of permanent lakes with little variation in inflows. Further, construction of dams and weirs has transformed flow regimes, resulting in habitat modification and altered wetland states (Kingsford 2000; Smith and Smith 2014). The categorisation of ‘temporary’ and ‘permanent’ is rigid and cannot take account of the variability of changes in flow and flood regimes.

Regarding adequacy of definition, within the aquatic system classes, the Basin ANAE classification uses attributes of water type and regime, vegetation and landform to classify aquatic ecosystems into specific types. However, these types are either too vague to provide useful information for conservation purposes (e.g. Psp2.1 Permanent salt marsh, Pt4.2 Temporary wetland, Pt4.1 Temporary floodplain wetland) or too specific, lacking justification of why they need to be distinguished from similar wetland types. For example, it is unclear whether there is a meaningful eco-hydrological difference between the Types Lt1.1 Temporary lakes and Lt1.2 Temporary lakes with aquatic beds (i.e. with submerged macrophytes) (Brooks et al. 2014, cf. p. 63). Both types share most features except for the presence of submerged macrophytes, which is an inconsistent and dynamic phenomenon. If lakes are temporary, submerged macrophytes dry out and die during dry periods. Using their presence as the primary indicator to differentiate lake types, without specifying other factors (e.g. water quality, salinity, water level), is unreliable, because their presence or absence may simply reflect different phases of the same system under varying hydrological conditions.

Regarding consistency, the Basin ANAE classification is inconsistent in distinguishing wetlands on the basis of representative features. Many types overlap because of the application of specific attribute rules applied to the same vegetation community. For example, wetlands containing river red gum as the dominant vegetation are split into four floodplain types (F1.1 Upland river red gum forest floodplain, F1.2 River red gum forest floodplain, F1.3 Upland river red gum woodland floodplain and F1.4 River red gum woodland floodplain) and two palustrine types (Pt1.1.1 Intermittent river red gum floodplain swamp and Pt1.1.2 Intermittent river red gum swamp). This splitting of categories for river red gum forest and woodland, which are widespread and easily recognisable vegetation communities with clear sets of water and habitat requirements (Roberts and Marston 2011), introduces a degree of unnecessary complexity and does not accord with ecological reality on the ground.

Under the ANAE classification, there are 15 ecosystem types for the Macquarie Marshes (Fig. 1a). However, some are not clearly defined and appear ecologically vague or meaningless from a conservation perspective. For example, the ANAE Type Pp4.2 Permanent wetland is described as ‘permanent wetlands not on floodplain with unspecified vegetation’ (Brooks et al. 2014, p. 66). Pt2.3.2 Freshwater meadow refers to temporary meadow on floodplain dominated by grasses and forbs with scattered shrubs, trees and sedges (Brooks et al. 2014, p. 69). These descriptions do not specify characteristic wetland plant species of each ecosystem type, making it difficult to determine ecological relationships. By contrast, the same area under the New South Wales classification (Fig. 1b) contains fewer but more ecologically meaningful groups (e.g. Inland Riverine Forests, Inland Floodplain Swamps, and Inland Floodplain Woodlands), that accurately represent wetland diversity in this region and the inter-relationship between vegetation communities and hydrological connectivity.

Fig. 1.

Comparison of wetland classifications for Macquarie Marshes: (a) ecosystem diversity using the Basin ANAE classification (Brooks 2024; Department of Climate Change, Energy, the Environment & Water 2021) and (b) vegetation classes using the New South Wales vegetation classification (Keith 2004; Benson 2006).

Regarding information quality, the Basin ANAE wetland types do not provide information at a scale useful for wetland conservation and management. Descriptions of vegetation communities are limited to a single structural form (grass, forb, tree, shrub) and species (coolibah, lignum, river red gum) or are too general to be of use (tall emergent aquatics, other aquatic trees, saltmarsh). Information on vegetation community composition, structure, and diagnostic species is lacking. This information gap ignores the co-occurrence of distinctive wetland plant species and communities within particular wetlands and their shared water requirements that reflect characteristics of water availability and flow regimes. This is a major limitation of the Basin ANAE classification. Furthermore, the use of the term ‘ecosystem type’ (Brooks 2022) is problematic. Ecosystems are generally defined as a set of biotic communities that interact with each other and their abiotic components in a discrete system of functions and processes controlled by common variables such as climate, water regime and soil type. Examples include temperate rainforests and alpine grasslands. But in the Basin ANAE classification, the term is used for every differentiable unit of the wetland landscape and includes entities that are small-scale habitat types or topographic features. Claypans are not ecosystems but are referred to as such in the Basin ANAE classification (Brooks 2022, cf. p. 21). This approach of splitting such units and calling them ecosystem types creates a false impression not only of wetland ecosystem diversity in the Basin but also implies that ecosystems are isolated and discrete rather than dynamic and strongly interconnected.

Regarding reproducibility, the ANAE is a strongly rules-based system. However, the underlying rationale for each class is not explained, and the descriptions of each ANAE ecosystem type are confined to the title only and thus lack clarity. Identification of these ecosystem types by independent users is therefore challenging without additional internal expert knowledge.

In summary, the Basin ANAE classification fails on the four criteria of adequacy of definition, consistency, information quality and reproducibility. It does not reflect distinctive habitat and vegetation characteristics that enable effective identification and inventory for conservation purposes or support international requirements under the Kunming–Montreal 30 by 30 goal.

State-based wetland classifications

Queensland has a two-level classification system of 98 higher-level BVGs with 1429 regional ecosystems that are qualitatively related to them (Neldner et al. 2023; Queensland Government 2024), of which 288 are wetlands, but only ~17 are contained within the Basin. REs are basically vegetation communities and include detailed descriptions of structure and species composition (Queensland Government 2024). For REs, there are three levels of organisation, namely, biogeographical regions, land zones and the vegetation communities within them (Addicott et al. 2021). Wetland REs are defined by satisfying one of the following criteria: the presence of hydrophytic vegetation, hydric soil or non-soil substratum (Environment Protection Authority 2005). However, the definition excludes floodplains that are flooded intermittently or episodically, which are crucial for some ecologically meaningful wetland vegetation in the basin such as coolabah and river cooba-dominated communities (Roberts and Marston 2011).

The Queensland classification contains clear definitions of regional ecosystems and BVGs. Regional ecosystem descriptions provide adequate floristic and habitat descriptions useful for determining important wetland characteristics. However, the classification is biased towards vegetation structure. Ecologically distinct wetlands with similar structure can thus be lumped into the same category. For example, BVG 16a open forests and woodlands includes river red gum-dominated forests and woodlands, river oak forests and coolabah-dominated woodlands, although their understorey communities and water regimes are completely different. Such a bias towards structure potentially makes conservation decisions more difficult and complex. Regarding consistency, the inclusion of bioregions, land zones and vegetation communities in the definition of REs results in unique classes with very limited representation outside of Queensland. Many REs overlap but are regarded as separate because they are in different bioregions. For example, RE 4.3.4, RE 6.3.7 and RE 11.3.3 are differentiated by bioregion (6, Mulga Lands; 11, Brigalow Belt; 4, Mitchell Grass Downs), but refer to the same wetland type, dominated by Eucalyptus coolabah woodlands on alluvial plains. The classification ranks low on reproducibility because RE definitions require expert procedures (Addicott et al. 2021).

The Victorian wetland classification (Department of Environment, Land, Water and Planning 2016) builds on previous classifications and inventory (Robertson and Fitzsimons 2004) and is based on ANAE principles, modified for Victorian environments. It includes regional and landscape attributes such as climate, landform, topography, hydrology and vegetation. Regional attributes are based on Victorian bioregions and subregions. Merged into this system is the broader Victorian vegetation classification based on EVCs. The wetland classification includes a category called wetland landscape (e.g. lowland grassy plains, lowland riparian floodplain) in which are nested ANAE wetland systems (lacustrine, palustrine, marine and estuarine). Within wetland systems are categories of vegetation as emergent or non-emergent, and the lowest level includes 159 EVCs used to classify wetland vegetation communities on the basis of floristics, ecological characteristics and life-forms (Department of Environment, Land, Water and Planning 2022; Department of Energy, Environment and Climate Action 2024a, 2024b). However, because wetland EVCs have not been fully mapped and there are many of them, they are grouped into seven larger categories (e.g. forest/woodland, sedge/grass/forb, shrub), which are equivalent to dominant vegetation categories in NVIS (Department of Environment, Land, Water and Planning 2016, cf. p. 31).

The Victorian system ranks relatively highly for adequacy of definition and information quality compared to other classification systems such as NVIS and ANAE. For wetland EVCs, indicator species and their associates are detailed, along with criteria to determine condition and extent of modification (Department of Environment, Land, Water and Planning 2022). Regarding consistency, the Victorian system is highly State-specific and quite complex because of the large number of wetland EVCs and its amalgamation of prior State-based wetland classifications with the ANAE classification and the State vegetation classification. Accordingly, there are similar issues with consistency as for the Basin ANAE aquatic ecosystem classes (cf. above). The Victorian system cannot be used to adequately represent wetlands at Basin scale because Victoria accounts for only 6% of the total wetland area of the Basin, and many of the wetland EVCs are not found outside Victoria. For reproducibility, the Victorian system ranks low. It is technically complex, requiring expert knowledge for effective use and, like the Basin ANAE classification, is strongly based on assignment rules, the rationale for which is not always clear.

The New South Wales system is a nested three-level vegetation classification to support conservation planning, policy and management (Office of Environment and Heritage 2018). The highest level, vegetation formation, is based on 12 groups distinguished by structural and functional features (e.g. arid shrublands, grassy woodlands, forested wetlands), with the next level, 99 vegetation classes, being nested within them (Keith 2004). Vegetation classes are defined by floristic, structural and functional similarities, representing broad groupings of related communities (e.g. for the Forested Wetlands formation, the classes are Coastal Swamp Forests, Coastal Floodplain Wetlands, Eastern Riverine Forests and Inland Riverine Forests). The lowest level is PCTs, which are assemblages of plants that live together and share habitat and environmental requirements (Benson et al. 2006). There are currently 1841 PCTs, of which 150 are for wetlands within the Basin (Department of Planning and Environment 2022).

In terms of adequacy of definition, each of the three major groups is well-defined and clear. Descriptions of vegetation classes include detailed information on ecology, topography, climate, hydrology, threatening processes, indicative species and environmental history, integrating and synthesising previous research (Keith 2004). The New South Wales classification provides a standardised system of inland wetland vegetation classes that can be applied consistently across the entire Basin. Information quality is high, enabling clear wetland delineation at a scale that can be used for conservation policy and management at an appropriate scale. And the system ranks high on reproducibility because it is simple, clear and well documented.

We considered the New South Wales classification was the most appropriate at Basin scale, enabling us to integrate plot-based vegetation records from each State into a unified system of vegetation classes to provide information to inform wetland conservation policy and management.

Distribution and extent of wetland vegetation classes

We added an additional class, Estuarine Wetlands for the Coorong and Lower Lakes of South Australia, to the eight wetland vegetation classes from the New South Wales classification. Because our focus is on wetlands of riparian zones and floodplains that are structured and maintained by particular flow and flood regimes, we excluded the classes Alpine Bogs and Fens, Montane Lakes and Montane Bogs and Fens as they fall outside the scope of environmental water management within the Basin.

The main characteristics of the wetland vegetation classes in the New South Wales classification (Fig. 2) are summarised in Table 2. Assigned to these classes were 52 Queensland REs (including subunits of REs), 55 Victorian EVCs, 71 New South Wales PCTs and 75 of the South Australian vegetation codes (cf. Supplementary Tables S1–S4). The total extent of the nine classes in the Basin is 106,337 km² (10% of its area; Table 3). Below, we outline the distribution and extent of these classes. Detailed descriptions of each are given by Keith (2004) and for Estuarine Wetlands by Phillips and Muller (2006).

Fig. 2.

Distribution of wetland vegetation classes in the Murray–Darling Basin on the basis of the New South Wales wetland vegetation classification (Keith 2004; Benson 2006). Layers for land use and protected areas measured using the geometry calculation in ArcGIS Pro. Protected areas are those located within wetlands. Intensive agricultural activities include horticulture, irrigated agriculture, dryland cropping and plantation forests.

North-west Floodplain Woodlands (Fig. 3a).

This is the most extensive class, covering 37,023 km² (35% of Basin wetlands) of mid-upper floodplains in northern New South Wales and southern Queensland, dominated by coolibah (Eucalyptus coolabah) with bimble box (E. populnea ssp. bimbil). Shrubs include river cooba, plus a variety of understorey grasses and sedges. It is closely allied with Inland Floodplain Woodlands, but tends to have a grassy understorey rather than saltbush. It grades into Semi-arid Floodplain Grasslands and Inland Floodplain Shrublands at higher elevation and lower flood frequency.

Fig. 3.

(a–d) Maps showing the distribution of individual wetland vegetation classes in Fig. 2. (e) Map shows wetland vegetation, agricultural land and protected areas (PAs). (f) Map shows wetland vegetation within and outside the actively managed floodplain. (b, d) Maps were re-sampled to a pixel size of 200 m for clearer visualisation. Other maps (a, c, e, f) are of 10-m pixel size.

Rationale for a vegetation-based classification

Differences in the distribution and availability of water in the landscape, and in the frequency and duration of periods of wetting and drying, define the ecological character of wetlands (Junk et al. 2014), as reflected in the eco-physiological adaptations of plants and their water requirements (Brock and Casanova 1997; Casanova 2011). Under the highly variable climatic and hydrological regime of the Basin, water is the most important environmental factor influencing vegetation structure and physiognomy, species composition and function (Keith 2004, cf. p. 15; Catford et al. 2017). Wetland vegetation underpins primary productivity, decomposition, nutrient and energy transfer, habitat and food provision for other organisms and modification of microclimate. Without vegetation, these ecosystem functions would not occur. It follows that a vegetation-based classification is based on the fundamental principles of ecosystem structure, function and character.

A vegetation-based classification that focusses only on floristics, or species assemblages, may be open to the criticism that it does not include important characteristics of climate, geomorphology, topography and other abiotic variables (Semeniuk and Semeniuk 1997; Robertson and Fitzsimons 2004). However, particular plant communities occur in places that share a particular set of abiotic conditions. In the Basin, the patterns of distribution and occurrence of plant communities reflect common climatic, topographical and hydrological habitats. To address the criticism that the vegetation-based Basin wetland classification does not include important abiotic variables, below we link it with the Whole-of-River Framework, which incorporates elements of flow regime, climate and topography (Roberts et al. 2016).

The Murray–Darling Basin wetland vegetation classification and the Whole-of-River Framework

The Whole-of-River Framework is a model of contrasting habitat types that riverine and wetland plants are adapted to, with particular adaptive traits of growth and reproduction relating to drought and salt tolerance, temperature extremes, flow regimes and water availability (Roberts et al. 2016). It consists of two gradients, flow energy and flood frequency, which effectively integrate variables for rainfall, evapotranspiration, temperature, flow regime and topography. Flow energy ranges from high to low and flood frequency from frequent to infrequent, creating four broad climatic-hydrological habitats (Table 5).

The wetland vegetation classes in Table 2 fit well with the Whole-of-River Framework, with most of the floodplain woodland classes in the low-energy, infrequently flooded group. The low-energy, frequently flooded group contains Inland Riverine Forests, Inland Floodplain Swamps and Estuarine Wetlands, which are the classes most commonly targeted for management with environmental flows. The high-energy, frequently flooded group contains only Eastern Riverine Forests of the river corridors of upper catchments, which are not amenable to management with environmental flows. The high-energy, infrequently flooded group contains Semi-arid Floodplain Grasslands and Inland Saline Lakes of the northern Basin.

Implications for environmental water management

Establishing protected areas for wetlands is insufficient for their conservation without effective management (Fitzsimons and Robertson 2005). In the Basin, maintaining or improving wetland condition depends on the delivery of environmental flows to priority sites (Berney and Hosking 2016). However, the implementation of environmental flows has resulted in ~80% of the volume being delivered as within-channel flows rather than as floods to wetlands (Chen et al. 2021). Where environmental water has been provided for wetlands, flows have tended to be delivered to easily accessible areas and floods were of limited extent and duration, often during seasons unlikely to result in growth and reproduction of wetland biota (Capon and Capon 2017; Chen et al. 2021). At Basin scale, environmental water requirements, i.e. the frequency, magnitude, duration and timing of flows needed to achieve environmental objectives, have not been met (Chen et al. 2021). This situation is likely to be a major contributing factor to the ongoing poor health of wetlands across most of the Basin (Sheldon et al. 2024). The process for prioritisation of wetlands for environmental flows is unclear, but appears to be based largely on which biotic components might be assumed to benefit from a particular type of flow event (Murray–Darling Basin Authority 2024c). The framework for determining Commonwealth environmental water use states the objective of maximising environmental benefits (Principle 3), but with no consideration of environmental water requirements of wetland biota, including vegetation communities (Commonwealth Environmental Water Office 2013).

Improvements in environmental flow decision-making, planning and implementation are required if these deficits are to be addressed effectively. Such improvements include policy options to address the need to adapt to less water in the future and re-frame the decision context from contestation between water for irrigation versus the environment towards water for adaptation under climate change (Colloff and Pittock 2022). We propose that this approach is more likely to be enabled by a wetland classification based on principles of ecosystem ecology, relatedness of vegetation communities and their environmental water requirements rather than, as for the Basin ANAE classification, on an artificial taxonomy based on fragmented landscape units of doubtful ecological significance and little or no predictive value regarding water requirements of vegetation communities.

Regarding wetland vegetation classes most likely to benefit from environmental flows, only Estuarine Wetlands and Inland Riverine Forests have more than 30% of their extent within the actively managed floodplain (Table 4). At least 80% of the extent of the other classes falls outside this area. Even if and when constraints on environmental water delivery are removed (Kahan et al. 2021), large areas of wetland vegetation, including those of high conservation value, will continue to depend on occasional high, unregulated, flooding flows for their maintenance and persistence. Assessment of the extent to which environmental flows can be delivered to representative areas of wetland vegetation remains an important consideration in wetland conservation policy and management under climate change.

Applications of the Basin wetland vegetation classification to conservation policy under climate change

Australia has committed to conserving 30% of ecologically representative terrestrial, inland water, coastal and marine ecosystems in protected areas by 2030 (United Nations Environment Programme 2022; Department of Climate Change, Energy, the Environment & Water 2024b). The extent of Basin wetland vegetation classes in protected areas is well below this target at 7% in Queensland, 6% in New South Wales and 8% overall (Table 3), although Victoria has achieved better implementation of a wetland reserve systems than have the other States (Nevill 2007; Pittock and Finlayson 2011). Some 22% of the Australian landmass is within the National Reserve System (Department of Climate Change, Energy, the Environment & Water 2023b) although there is a lack of clarity in national conservation policy on what constitutes ecologically representative ecosystem types. There is no definition of wetland ecosystems in the Interim Biogeographic Regionalisation for Australia (IBRA) which is the current tool for prioritising land conservation. The under-representation of wetlands in protected areas seems, at least in part, to relate to the prioritisation of terrestrial and marine ecosystems. In the draft national roadmap for protecting and conserving 30 by 30 there is no specific mention of any strategy to protect wetlands and rivers (Interjurisdictional Working Group 2024).

To meet the 30% target, more effort is needed to protect representative areas of wetlands that align with the CARE principles for conservation planning and policy (Possingham et al. 2006). Wetland classification characteristics have a major effect in determining conservation status of freshwater wetlands (Robertson and Fitzsimons 2004). The Australian government is considering the use of an appropriate classification to help implement the Global Biodiversity Framework, which includes the IUCN global ecosystem typology (Keith et al. 2022). The Basin wetland vegetation classification presented herein can be appropriately aligned with the ecosystem functional groups in the IUCN typology (Table 2).

The Basin wetland vegetation classification provides a simple, practical basis for determining representativeness for conservation of wetlands in protected areas. The classification shows the distribution and extent of nine distinctive wetland vegetation classes, together with their abiotic characteristics within the Whole-of-River Framework. Each wetland vegetation class has been defined in relation to its location within the landscape, its current and future climate conditions, its water requirements and the flow regime it is currently subjected to, as well as its amenability to conservation and management with environmental flows. These attributes form a sound basis for conservation decision-making now and into the future.

Changes in wetland extent, character and condition are likely to increase with less water available under climate change and increased demand from consumptive users. A major advantage of the Basin wetland vegetation classification is that classes are defined in ways that reflect their relatedness regarding patterns of change in driver variables. For example, Inland Riverine Forests merge into, and are replaced by, Inland Floodplain Woodlands and North-west Floodplain Woodlands, with increasing elevation from the main river channel and decreasing rainfall, flood frequency and duration. Inland Floodplain Shrublands merge into Inland Floodplain Swamps under increased flood frequency and extent and into Inland Floodplain Woodlands as the frequency and extent decrease (Keith 2004, cf. pp. 218, 250).

Our revised classification has strong predictive value for assessing the persistence or transformation of wetland classes under different future climate scenarios. This property is essential for adapting wetland conservation under climate change (Schweizer et al. 2022). A robust classification that considers spatio-temporal hydrological change and its impact on wetlands is crucial for tracking alteration in wetland ecological character (Kingsford et al. 2021). It allows us to define boundaries of representative wetlands and predict how changes in flow and flood regimes are likely to influence ecosystem structure, function and species assemblages. This knowledge enables the establishment of clearly defined baselines for limits of acceptable change in ecological character (Rogers et al. 2013), which i a fundamental requirement under the Ramsar Convention (Pittock and Finlayson 2011; Finlayson et al. 2017), but one which is not possible to achieve using the Basin ANAE classification.

Changes in ecological character owing to altered flow regimes caused by irrigation diversions and climate change have already occurred in wetlands in the Basin. For example, at Barmah Forest, the invasion of giant rush (Juncus ingens) and river red gum into spiny mud grass (Pseudoraphis spinescens) plains (Bren 1992; Colloff et al. 2014; Vivian et al. 2014). In parts of the Macquarie Marshes, Inland Floodplain Swamps have transformed to Inland Floodplain Shrublands as a result of major alterations in the flow regime owing to water diversions and the difficulties of maintaining wetlands with environmental water only (Berney and Hosking 2016; Bowen et al. 2019).

Classification and mapping provide information on the extent and boundaries of each wetland and how these change over time. Current extent can be compared with estimated pre-European extent to determine the effects of historical drivers of change and threatening processes. If wetlands in protected areas are to be managed effectively to mitigate adverse effects from climate change, then monitoring ecosystem change needs be conducted on a regular basis. At Basin scale, this process is enabled by a simple, predictive wetland vegetation classification.