An Estuary Ecosystem Classification that encompasses biogeography and a high diversity of types in support of protection and management

For nearly three decades, the Whitfield (1992) characterisation scheme served as a reference framework to type South African estuaries. We outline a revised ecosystem classification scheme that incorporates biogeographical zonation and introduces new types. Coastal outlets were re-categorised as estuaries or micro-systems. For functional estuaries, the Estuarine Lakes, Estuarine Bays and Predominantly Open Estuary types were largely retained. New types are Estuarine Lagoons and Arid Predominantly Closed Estuaries. The numerically dominant, temporarily open/closed category was subdivided into Small and Large Temporarily Closed Estuaries, with a total habitat area of 15 ha, serving as threshold separating these two subdivisions. River mouths were renamed Fluvially Dominated Estuaries and divided into large and small size categories to reflect dissimilar catchment influences. Micro-systems were separated into micro-estuaries, micro-outlets, and waterfalls. South Africa’s 290 estuaries were classified into 22 estuarine ecosystem categories arising from nine estuary types occurring across four biogeographical zones. In addition, 202 micro-systems were classified into nine ecosystem types, of which only the micro-estuaries (42) share possible functionality with estuaries. Estuaries subjected to functional shifts were also identified. The classification system provides a framework that integrates biogeography and the range of biophysical parameters evident in South Africa, and can be used for red listing of ecosystem types and determining estuarine sensitivity to pressures.

Estuaries are difficult to classify, because they vary temporally in shape and size and encompass gradients in conditions from riverine to marine. Human interventions and morphological changes brought about by climate and sea level fluctuations further complicate the process. This paper recognises an estuary as a partially enclosed permanent waterbody that is either continuously or periodically open to the sea and that extends as far as the upper limit of tidal action, salinity penetration or back-flooding under closed mouth conditions. During high catchment flows or floods, an estuary can become a river mouth, with no seawater entering the formerly estuarine area or, when there is little or no fluvial input, an estuary can be isolated from the sea by a sandbar and become fresh or even hypersaline (CSIR 1992; modified from van Niekerk and Turpie 2012).
"Typology" refers here to the characterisation of estuary types according to shared key features, whereas "ecosystem classification" refers to the categorising of estuarine ecosystem types based on their abiotic constituents (e.g. climate, oceanic conditions, substrate, water, and all other non-living elements) and its biotic constituents, consisting of all its living members (Allee et al. 2000). Ecosystem classifications are complex, and often hierarchical or nested, whereas typologies tend to be more straightforward and based on general type. The level of classification depends mainly upon the number of criteria selected (e.g. biogeography, geomorphology and biology) and the spatial resolution required (e.g. local, regional or global). The greater the number of criteria and the wider the geographical scope, the more complex the classification

An Estuary Ecosystem Classification that encompasses biogeography and a high diversity of types in support of protection and management Introduction
Open Access article distributed in terms of the Creative Commons Attribution License [CC BY 4.0] (http://creativecommons.org/licenses/by/4.0) becomes. Depending on the task or information required, a balance has to be established between the number of criteria and the level of detail selected (Whitfield and Elliott 2011).
Reflecting the high diversity of estuarine types globally, numerous estuary typology and classification schemes exist. These are usually specific to a geographic region (Whitfield and Elliott 2011). Traditionally estuaries have been typed based on key processes and features, such as tidal ranges, tidal prisms, topography, geomorphology, salinity characteristics and ecosystem energetics (Davies 1964;Nichols and Biggs 1985;Kennish 1986). Topographical typologies categorise them as drowned river valleys, fjords, bar-built estuaries and others (Pritchard 1952;Dyer 1997), whereas morphological typologies base groupings on physical features resulting from the interplay between catchment runoff and sediment loads, and tides, waves and other coastal processes (Dalrymple et al. 1992). Salinity-based approaches separate them according to the degree of mixing within the water column (Pritchard 1955;Cameron and Pritchard 1963) and stratification-circulation typologies using densiometric numbers from the fluid mechanics discipline (Hansen and Rattray 1966;Fischer 1972;Simpson et al. 1990). Most of these typologies require extensive in-field time-series and high spatial coverage data. National or regional ecosystem-level classification schemes have to recognise environmental parameters often not strongly reflected in physical and morphology typologies, such as variations in climate or biogeography, vegetation and other biological aspects (see Whitfield and Elliott 2011 for the evolution of estuary classification). Examples include the comprehensive United States marine and estuarine ecosystem and habitat classification that moves from a biogeographical level to a habitat level at it smallest spatial scale (Allee et al. 2000). Regional schemes that explicitly include temporarily closed estuaries, common on the South African coastline, are those of Australia and California. Early Australian schemes identified seven geomorphology-based estuary types across five biogeographical regions under the influence of wave, tide and river energy (Boyd et al. 1992;Dalrymple et al. 1992;Kench 1999). Linking these to climate and rainfall characteristics allowed for the incorporation of freshwater-and evaporation-dominated types (Boyd et al. 1992;Heggie et al. 1999aHeggie et al. , 1999bKench 1999). A more recent Australian classification described three intermittently open/closed estuary types (IOCE) based on the duration and frequency of mouth condition and estuary size (large, medium or small) (McSweeney et al. 2017). A typology of Californian estuaries distinguished eight closed-mouth states based on berm elevation and tidal exchange where river inflow rather than tidal influence controls mouth opening. This scheme recognised that mouth states prevail over multiyear to multidecadal periods (Jacobs et al. 2010). Over the past six decades estuary classification studies have therefore evolved from relatively simple "topology" schemes (defining estuary types based on a key process or feature), to more complex regional ecosystem-level classification schemes (regional schemes that include elements of climatic/biogeography, estuarine processes and biological responses).
The choice of classification system depends to some extent on its intended purpose. Ecosystem-level classification schemes have a wide range of potential application in research and natural resource management. National and regional ecosystem classifications are useful, because they can be applied to the International Union for Conservation of Nature (IUCN) Red List of Ecosystems (RLE) framework to assess risks and identify vulnerable ecosystems based on a set of criteria and thresholds (Rodríguez et al. 2011;Keith et al. 2013;2015;Bland et al. 2017aBland et al. , 2017bBland et al. , 2018.The RLE requires clearly defined units of assessment (ecosystem types) that can be spatially delineated (Keith et al. 2013). Ecosystem classification is therefore one of the fundamental inputs for such global assessment processes.
South African national-level biodiversity assessments, and national and regional conservations plans also require an ecosystem-level classification that details similarities and differences between estuary types and describes their biogeographical occurrence (Turpie and Clark 2007;Turpie et al. 2012;van Niekerk and Turpie 2012). Classification is fundamental to estuary conservation planning to pinpoint important biodiversity areas and initiate the setting of targets for species, habitats and types to ensure that all species and the critical processes they depend on are conserved (Turpie and Clark 2007;Turpie et al. 2012). It is also essential for determining estuary flow requirements and water resource allocations where estuary type serves as an indicator of sensitivity to flow modification and declining water quality (van Niekerk et al. 2019a(van Niekerk et al. , 2019b. In a data-limited environment, such as South Africa, classifications can be used to signify system-specific ecosystem processes and associated biotic characteristics.
The aim of this paper was to develop an ecosystemlevel national classification scheme for South African estuaries that incorporates key processes and patterns in different biogeographic regions and can inform broad-scale assessments of estuary resilience to anthropogenic pressures.

Material and methods
A detailed literature review of globally-(see Introduction) and locally-used estuary typologies was conducted to identify key parameters that support estuary classification at the ecosystem level. Parameters used most often in estuary typologies included: estuary size and geomorphology; catchment features (size and geology); river flow and tidal regimes; marine connectivity (mouth state or inlet stability); sediment processes; salinity structure; habitat diversity; and, biotic composition. In addition, common to most national or global ecosystemlevel classification schemes were biogeographical, ocean processes (wave energy), or climatic elements. Developing an estuarine ecosystem-level classification scheme for South Africa required a number of steps ( Figure 1): revisiting the biogeographic organisation of South African estuaries to ensure that all climatic and oceanic processes are accurately reflected based on emerging research; evaluating existing typologies and revisiting, if needed, the high diversity of estuarine processes and responses that occur on the subcontinent. In addition, previously neglected very small outlet types that do not fully function as estuaries, collectively referred to here as micro-systems, were included. A "functional" estuarine system, even if very small (<2 ha or <200 m in length), was taken as any permanent coastal waterbody that supports significant estuarine processes and associated biological composition.
A significant effort was made to ensure all rivers and streams with outlets to the sea were identified and to address systems omitted from historical lists (>400 outlets were evaluated). Datasets consulted included the CSIR NRIO list (1981) and those of Harrison et al. (2000), van Niekerk and Turpie (2012), Whitfield and Baliwe (2013), Bate et al. (2017), Dalu et al. (2018) and Human et al. (2018). Micro-system types, previously omitted from formal classification schemes, were also identified through the literature and using satellite imagery (GoogleEarth TM ) and aerial photographs (1930s to 2000s). Where possible, a distinction was made between natural outlets and artificially created stormwater outlets (with the latter validated by site visits), with a focus on only including naturally occurring features in the landscape. Seven coastal outlets, not previously listed as estuaries, were deemed to be functional system types based on size, topography and vegetation cover and added to the formal list of estuaries. All larger outlets (>200 m) were delineated based on the 5 m topographical contour (obtained from the Department of Rural Development and Land Reform: National Geospatial Information) to determine how far they extend inland van Niekerk et al. 2019c). Existing national and bioregional datasets were then scrutinised to establish what information was available on the physical features and processes of estuaries (e.g. size, mean annual runoff, mouth state) and biotic components (plants, invertebrates, fish and birds) to assist with classification (Harrison et al. (2000); van Niekerk and van Niekerk et al. 2015;van Niekerk et al. 2019d).
Using the collated information, all functional estuaries were evaluated by a panel of national experts in 2009 and again in 2018 (see author list), each expert having between 10 and 35 years' experience in estuarine physical processes, vegetation, invertebrates and/or fish (van Niekerk et al. 2013). The specialists assessed each dataset to identify estuary types that support key ecosystem processes and functions across biotic groups. Although published information was given preference, the panel also drew on the authors unpublished field data and personal field observations. The outcome was cross-validated by consulting historical data sources and published literature. Data and field observations were often lacking for very small estuaries and outlets, and classification for these systems relied on visual observations and expert judgements. Usually the panel participants' consensus view on a system's classification was adopted, with the few exceptions being resolved by a majority vote.

South Africa's estuarine ecosystem-level classification
Currently, there are a number of typologies or classification systems recognised for estuaries in South Africa, largely based on geomorphological and/or physical characteristics. Early attempts described estuary types fairly loosely and with little scientific backing, e.g. temporarily closed estuaries were referred to as "lagoons" (Begg 1978), "blind estuaries" (Day 1980(Day , 1981 or "seasonally open/ closed estuaries" (Bennet1989). Whitfield (1992) typed South Africa's estuaries into five types based on their physiographic (tidal prism, size), hydrographic (mouth state and mixing process) and salinity characteristics (Table 1). A geomorphological system used by Harrison et al. (2000) recognised six main types based on mouth condition (open or closed), size and the presence of a sand bar at the mouth (or inlet Does it support estuarine functionality and/or biota? Figure 1: Flow diagram of estuary classification process (van Niekerk and Turpie 2012) proposed a typing based on estuary size, mouth state, salinity structure and catchment type. This resulted in 46 ecosystem types, which included three estuary size classes (large >100 ha, medium 10-100 ha and small <10 ha). Whitfield's (1992) typology of estuaries has been most widely used to describe the estuaries on the subcontinent (Table 1). Most estuaries were classed as Temporarily Open/Closed systems, where connection to the sea is governed by a sand bar that builds across the mouth under conditions of low river flow and high coastal wave conditions. Floods cause the mouth to open by scouring large volumes of sediment to sea. Permanently open estuary types are controlled by marine and riverine processes that act together to maintain an open mouth. Estuaries categorised as river mouths are dominated by riverine processes, because strong outflow inhibits marine intrusion into the mouth area, allowing oligohaline conditions to persist, especially in the upper and middle reaches. Heavy silt loads are a common feature of river mouths, resulting in shallow estuaries and ebb tidal deltas. Estuarine Bays and Estuarine Lakes, the two remaining categories, are large open water systems uncommonly represented along the coast. The bays are permanently linked to the sea by deep mouths and have euhaline salinities in the lower and middle reaches whereas the lakes are usually drowned coastal systems filled by reworked sediments and separated from the sea by vegetated sand dunes. Estuarine Lakes can be either permanently or temporarily linked to the sea.
However, Whitfield's (1992) scheme omitted a large number of small estuaries and micro-systems from its original characterisation. The scheme also primarily covered functional estuary types and ignored regional biogeography. Atypical estuary types, such as coastal lagoons, where a river or stream is absent and where the groundwater provides only a limited freshwater signal, were also not represented. The scheme showed a bias towards estuaries important to fish, and therefore largely ignored the small, hypersaline systems along the west coast, which have few fish, but are important for plants and birds. Nearly 75% of all estuaries were classified as temporarily open/closed systems, and the scheme did not recognise the differences between the more habitat diverse/species-rich larger systems, and the relatively depauperate smaller examples of this type. Finally, the river mouth type encapsulated a diversity of fluvially dominated estuary types ranging from those being fed by large, sediment rich catchments to clear, smaller catchment, black-water systems with very limited suspended sediments.
The above limitations of the Whitfield (1992) estuarine classification necessitated the development of an expanded ecosystem-level classification scheme for South Africa's estuaries, although at the same time allowing for "cross walking" of historical studies and assessments. Developing a more comprehensive ecosystem-level classification scheme required three interlinked steps: 1) the formal biogeographic organisation of South African estuaries to reflect regional biotic responses; 2) the refinement of estuary types to reflect the high diversity of small and large estuary types that occur on the subcontinent; and 3) the inclusion and organisation of previously neglected micro-system types.

A new tropical biogeographic zone
The traditional biogeographic organisation of South African estuaries has included only three regions, viz. Subtropical, Warm Temperate and Cool Temperate (Emanuel et al. 1992;Turpie et al. 2000;Harrison 2002). However, the tropical distribution of some species recorded in the Kosi and uMgobezeleni Estuaries in the north, provides good motivation to subdivide the Subtropical region and include a tropical transition zone in the north-east of South Africa. Both of these systems, albeit differing in size, type and function, are unique clear water estuaries on a small section of tropical South African coast; the location also being characterised by the higher latitudinal limit of Western Indian Ocean coral reefs (Schleyer et al. 2008).
By way of example, the Kosi Estuary has a uniquely tropical flora amongst South Africa estuaries, indicated by presence of the seagrass Halodule universis and mangrove species Lumnitzera racemosa, Ceriops tagal and Xylocarpus granatum that are found in no other South African systems (DWS 2016). Recent records of tropical species, such as the snake sea cucumber, Synapta maculata, and pen shell, Pinna muricata in the Kosi Estuary point to a tropical influence on the system's invertebrate fauna (DWS 2016). The Kosi ichthyofauna comprises a particularly wide diversity of fish, including tropical species not reported from any other South African estuaries (Blaber and Cyrus 1981;DWS 2016). Although the presence of reef habitat at the mouth contributes to this, biogeographic considerations also play a role. Elements of the estuary-associated fish fauna in Kosi appear to reflect the unique (in a South African context) linkages between estuarine habitats, particularly clear water mangroves, and offshore coral reefs. This is evidenced by the abundance and large sizes of several members of the Lutjanidae (snappers) in the Kosi system. This family of fish includes many species that rely on linkages and strong connectivity between mangroves and coral reef habitats on tropical coastlines elsewhere in the world (e.g. Nagelkerken et al. 2000Nagelkerken et al. , 2002Mumby et al. 2004).
In contrast, to the south, St Lucia Estuary supports invertebrate and fish assemblages common in Subtropical estuaries on the subcontinent (MacKay et al. 2010;Perissinotto et al. 2013). The new zone also aligns with marine biogeographic patterns evident for floral and faunal groups, such as benthic macroalgae (Bolton et al. 2004), intertidal invertebrates and algae (Sink et al. 2005), and shelf holothuroideans (Thandar 2015), and agrees with the recent marine biogeographic map that includes the Natal-Delagoa tropical region (Sink et al. 2019).
Therefore, the updated ecosystem-level classification scheme defined the biogeographical regions as the tropical from Kosi to uMgobezeleni, the Subtropical stretches from the St Lucia system in KwaZulu-Natal to the Mbashe Estuary in the Eastern Cape, the Warm Temperate from the Mendwana Estuary in the Eastern Cape to the Ratel Estuary near Cape Agulhas, and the Cool Temperate from the Uilkraals Estuary to the Orange Estuary on the Northern Cape coast (Emanuel et al. 1992;Turpie et al. 2000;Harrison 2002).

Revision of South Africa's estuarine types
Improved scientific understanding has resulted in necessary updates to Whitfield's (1992) classification, e.g. adding two new estuary types, namely Estuarine Lagoon and Arid Predominantly Closed Estuaries. Langebaan is an example of an Estuarine Lagoon, whereas the Buffels, Swartlintjies and Spoeg Estuaries in the same region are examples of Arid Predominantly Closed Estuaries (Figures 2, 3 and 4, Supplementary material). The original "River Mouth" type was renamed "Fluvially Dominated" and divided into large and small systems, and according to river sediment inputs. The temporarily open/closed group was also divided by size into Large and Small Temporarily Closed Estuaries. The new types are further described in Figure 2 and Table 2, with key features and dominant physical processes that characterise these revised estuary types being summarised in Table 3. The listing and designation of all systems is provided in the supplementary material (Supplementary Table). Localities around the coast are shown in Figure 4.

Estuarine Lakes
Estuarine Lakes (also known as Intermittently Closed and Open Lakes and Lagoons or Coastal Lakes) comprise one or more typically large circular waterbodies connected to the sea by a constricted inlet channel ( Figure 3, Table 3). Freshwater input can be from a single or multiple large rivers, groundwater or aquifers, or multiple small waterways or streams feeding into the basin; or a combination thereof. Maximum water levels are determined by berm height, mouth state and freshwater input. Marine connectivity varies from almost permanently open, to temporarily closed, on annual scales. Salinities are highly variable, ranging from fresh to hypersaline, because of differing freshwater input (surface and ground water), evaporation and the extent and duration of the marine connection. Mixing processes are dominated by wind and, to a lesser extent fluvial inputs, owing to their restricted mouths and relatively   Figure 2: The extended South African estuarine typology into nine estuary types and three micro-system types large surface areas. Average tidal amplitudes are negligible (15-20 cm) when connected to the sea, primarily as a result of restricted mouth conditions (Department of Water and Sanitation tidal recorders G4T004, G4T003, K3T006, W3T002, W7T003). Sediment processes tend to be stable, with infilling occurring over long time scales and system resetting confined to larger flood events.

Estuarine Bays
Estuarine Bays (also known as coastal or estuarine embayments) are permanently linked to the sea by unrestricted, deep mouths and are dominated by tidal processes, with tidal amplitude close to that of the sea (Figure 3, Table 3) (Department of Water and Sanitation tidal recorder K5T001). These are large systems (>1 200 ha), with generally round basins, where only the upper reaches experience a degree of constriction to tidal flows. a result of relatively low river inputs they have a predominantly euhaline salinity regime in the lower and mid reaches, with freshwater mixing processes being mostly confined to the more restricted upper areas. Sediments are typically marine in origin and grain size distributions are stable over time. There are two natural occurrences of Estuarine Bays in South Africa (namely Knysna and Durban Bay; Figures 3 and 4).

Estuarine Lagoons
Langebaan Lagoon (Figure 3) has many of the characteristics of an estuary (Whitfield 2005), including calm waters that are protected from marine wave action and biota that reflect many of the species usually found in estuaries (Figure 4, Table 3). However, despite groundwater seeps into some areas, it lacks riverine inflow and a normal estuarine salinity gradient (Table 2). Langebaan Lagoon represents a unique coastal ecosystem type (Table 4). It is recognised as an estuary, because its ecological functioning includes both freshwater and marine inputs into a semi-enclosed embayment. Estuarine Lagoons, as defined here, are permanently connected to the sea and are therefore marine dominated. Tidal action is the dominant mixing process and sedimentary processes are therefore generally stable. Tidal amplitude and water levels are close to those of the sea.  Table 3). Tidal amplitude ranges from 0.75 to 1.5 m (Department of Water and Sanitation tidal recorders E2T014, G1T074). Predominantly Open Estuaries are linear systems where mixing processes are dominated by both fluvial inputs and tidal action creating vertical and horizontal salinity gradients. Under low river flows and high summer evaporation, hypersalinity can develop in the upper reaches. The degree to which the mouth is restricted depends on the rate and volume of freshwater inflow. Some systems become severely constricted during low flow periods, decreasing the tidal amplitude and increasing the duration of the ebb tidal cycle. Regular flooding results in relatively mobile sediments. These estuaries usually support wetlands, salt marshes, macrophyte beds and marine and estuarine fauna (Whitfield 1992). Surprisingly, their size varies considerably ranging from 10 to 7 500 ha, with smaller systems afforded a degree of protection against direct wave action by rocky headlands or subtidal reefs, which assists in maintaining an open mouth.

Large and Small Temporarily Closed Estuaries
The sizeable temporarily open/closed category of Whitfield (1992)      of open water area) as the dividing threshold (Tables 2  and 3). The division was based on differences in recorded biophysical processes and patterns (Teske and Wooldridge 2001;Vorwerk et al. 2003;James and Harrison 2017). Small systems are likely to experience rapid increases and decreases freshwater runoff over a few hours making them strongly event driven. There will be little scouring following berm breaching, and a semi-closed mouth condition can easily develop owing to a small, perched, outflow channel that restricts tidal amplitude to 10 to 15 cm (Department of Water and Sanitation tidal recorder T7T004). There is minimal water column area during the open state. Habitat diversity is likely to be low (and without intertidal salt marshes) and species diversity and abundance reduced. Significant differences have been noted between invertebrate and fish assemblages of Small and Large Temporarily Closed Estuaries along the southeast coast (e.g. Teske and Wooldridge 2001;Vorwerk et al. 2003;. Both large and small systems tend to be linear or funnel shaped, with highly restricted inlets. Smaller systems especially tend to be "perched" above normal tidal levels, resulting in little to no open water area during the open mouth low tide state. Water levels are dominated by the state of the mouth, with highest levels of 1 m to 2 m above mean sea level during the closed phase. Tidal ranges are very restricted, varying from 25 to 50 cm in Large Temporarily Closed Estuaries (Department of Water and Sanitation tidal recorders K2T004, K4H100, P4T002) to 15 to 30 cm in Small Temporarily Closed Estuaries (Department of Water and Sanitation tidal recorders U7T001, K8T004; unpublished field observations). Open phase mixing processes are dominated by fluvial input and partially by tides. When closed, wind and seepage losses through the berm play a key role. Sediment composition is largely stable, resetting mainly during floods. Salinity regimes range from almost fresh to hypersaline, which in large systems can develop during times of low flow or droughts. Small Temporarily Closed Estuaries tend to be fresher in character, because they have less connectivity with the sea.

Small and Large Fluvially Dominated Estuaries
Estuaries characterised as river mouths by Whitfield (1992) (also known as river-dominated and tidal river mouths) were divided into two categories, Small and Large Fluvially Dominated systems to distinguish between small (<15 ha), black-water dominated, rocky, temperate southern coast estuaries, and large, shallow, sediment rich, freshwater dominated systems of the east and west coasts (Figure 2, Tables 2 and 3). The larger systems have very high sediment turnover, often develop ebb-tidal deltas, are turbid and can close during periods of low flow, e.g. uThukela and Orange Estuaries (Figure 3). Large Fluvially Dominated Estuaries tend to be constricted and can even periodically close during low flows (Department of Water and Sanitation tidal recorders S7T008, T3T018, V5T003). Small, sediment-starved, fluvially dominated systems have unrestricted mouths, because they usually occur along rocky shores and receive clear humic-stained water from Table Mountain Sandstone catchments. Fluvial processes are dominant and salinities are mostly fresh throughout the estuary for more than half the time. During peak flood conditions, outflows can influence salinities for a considerable distance offshore.

Arid Predominantly Closed Estuaries
This type comprises six small estuaries, namely the Buffels, Swartlintjies, Spoeg, Groen, Sout (Noord) and Wadrift Estuaries, located in the Namaqua west coast region (Figures 3 and 4, Supplementary table). They are linear or funnel shaped and closed on annual to decadal time scales. Salinities tend to be euhaline to hypersaline, as a result of low fluvial input and high evaporation rates (Table 3). Accordingly, mixing processes tend to occur over long time periods and they are dominated by the effects of evaporation, winds and seepage through the berm at the mouth. Occasional breaching and overwash during high sea conditions provide for marine input and connectivity. Sediment processes are generally stable on decadal time scales and are reset by large intermittent flash floods. Water levels are determined by the interplay between sand berm level, evaporation rates and seepage losses. Groundwater and inflows from local fountains replenish these losses and influence the salinity regimes of these estuaries.
Arid conditions promote the growth of unique vegetation, such as salt tolerant, succulent Sarcocornia spp. and Salicornia spp. that can occur kilometres inland, making  . Fish diversity, abundance and community structure relies on "suicidal" recruitment that is largely a function of connectivity with the sea and the degree of overwash during high seas. Fish survival depends mostly on groundwater inflow maintaining a suitable salinity gradient, with at least some areas having a salinity not exceeding 50. Safe return of fish to the sea can occur during river flood events and depends on a quick breach and fish not suffocating in sediment-laden water backing up against the berm. Invertebrate diversity, abundance and community structure are related to changing salinity gradients, including long cycles of hypersalinity. The Swartlintjies, Sout and Groen Estuaries are hypersaline, with a high biomass of brine shrimp Artemia sp. that hatch at salinities >40 and encyst, sinking to the bottom when salinities exceed 150. Cycles of Artemia abundance follow salinity regimes that in turn affect the diversity, abundance and occurrence of flamingos and other birds that feed on them.

Micro-system types
The approximately 400 river and stream outlets along the coast were categorised into two broad categories: fully functional estuaries (described in Section 3.1.2) and micro-systems ( Figure 5), with micro-systems generally being permanent or non-permanent coastal waterbodies < 2 ha in area or < 200 m in length (Supplementary table). Historical datasets contain information on some outlets not deemed to be functional estuaries (Harrison et al. 2000). Nine small systems previously classified as estuaries were reclassified as micro-systems under the new classification ( Table 2), because of clear indications that they do not support typical estuarine functionality, e.g. some systems have limited estuarine biota (Magoro et al. 2019), with no fish fauna recorded in micro-outlets over a number of visits (Magoro 2018). Micro-systems were classified into three categories based on spatial features (e.g. size and length) and field biotic observations:

Micro-estuaries
These are defined as small, permanent coastal waterbodies where mixing of salt-and freshwater can periodically occur owing to overwash from the sea or tidal exchange following breaching of the mouth. These small systems are likely to support low densities of a limited number of estuarine and marine species (Bate et al. 2017;Dalu et al. 2018;Human et al. 2018;Magoro et al. 2019).

Micro-outlets
These are very small waterbodies (<1 ha in area or <50 m in length) that are ephemeral in nature (i.e. they can dry out during periods of low flow) or are elevated above mean sea level, with a perched outflow channel that does not facilitate tidal mixing of salt-and freshwater (Dalu et al. 2018;Human et al. 2018). It can, however, act as a limited conduit between the land and the sea during periods of elevated stream outflow or exceptionally high storm sea events.

Coastal waterfalls
This outlet type is represented by waterbodies elevated more than 10 m above mean sea level that have no direct channel connection with the sea. Because of their elevation, they do not serve as conduits between the land and the sea. These systems occur along rocky shorelines where the presence of bedrock does not allow for channel erosion to mean sea level. However, the continuous outflow of freshwater into rocky coastal habitats could support unique marine biotic assemblages along the coast. The localities of designated micro-estuaries, coastal outlets/seeps and coastal waterfalls are shown in Figure 5, including distributions across the revised biogeographical regions. Listing and designation of all systems is provided in the supplementary material (Supplementary table).

Developing an ecosystem-level classification scheme
The final step in the ecosystem classification of South Africa's 290 estuaries is to intersect the four biogeographical zones, namely the Cool Temperate (Orange to Ratel), Warm Temperate (Heuningnes to Mendwana), Subtropical (Mbashe to St Lucia) and Tropical (uMgobezeleni to Kosi) zones with the nine primary estuary types. This resulted in 22 estuary ecosystem categories for South Africa (Table 4) and represents a high diversity of estuary ecosystem types, which is not unexpected considering the country's diverse climatic, oceanographic and geological drivers. Overall the Tropical biogeographical region has the least number of estuaries, whereas the Subtropical (130) and Warm Temperate (124)  An additional 202 micro-systems were recorded around the coast, albeit with a very low confidence, because of lack of data (Table 5). Of these, a total of 42 micro-estuaries were tentatively identified, with five systems located the Cool Temperate zone, 13 in the Warm Temperate zone and 24 in the Subtropical biogeographical region. Micro-outlets and waterfalls were only identified in as much as that act as a potential conduit for land-sea interactions and should not be targeted for coastal and water resource development to preserve estuarine ecosystem services.

Estuaries subjected to functional type shifts and causative factors
Estuaries and coastal ecosystems are facing increasing anthropogenic pressures affecting their productivity and ability to provide ecosystem services (Borja et al. 2016). The main pressures on South African estuaries include flow modification, catchment degradation, coastal development, pollution, exploitation of living and non-living resources, and mouth (inlet) manipulation (van Niekerk et al. 2013). These have altered key defining features or processes (i.e. mouth behaviour, water level fluctuations, mixing processes) in some estuaries to the extent that they no longer function in accordance with their natural type (see Table 6 for summary of estuaries that were subjected to functional shifts). Type shifts should not be confused with degraded estuaries that still function in a similar way to natural, e.g. iSiphingo Estuary is severely degraded (i.e. has undergone a condition shift), as a result of pollution and habitat loss, but it still maintains its function as a Predominantly Open Estuary through a concrete pipe constructed in the sand berm at the mouth.

Estuarine lakes
The Zeekoei, Heuninges, uMhlathuze (Richards Bay) and iNhlabane systems were once Estuarine Lakes that have all been irreversibly altered through mouth manipulation and development ( Table 6). The Zeekoei lake system naturally closed for long periods until the 1940s when the connection between the lakes and main estuary channel was severed by weirs and levees that were constructed to prevent flooding of surrounding urbanised areas, while maintaining high water levels in the main waterbodies. A concrete canal currently connects the lake system to the sea. Although there is no tidal exchange, the system is essentially permanently open, but no longer functions as an Estuarine Lake.
Development and agriculture on the Heuningnes Estuary floodplain have contributed to the system function changing from an Estuarine Lake that used to close periodically, into a Permanently Open Estuary. More than a century ago, drains were excavated throughout the floodplain and adjacent wetlands to "reclaim" land for grazing and dry-land agriculture. Since the 1940s, sand dunes at the mouth have been manipulated to prevent mouth closure and natural inundation of the estuarine floodplain, much of which is now under farmland. Subsequently, the mouth has closed on only a few occasions. Artificial breaching is then practised to ensure an open mouth. The topography suggests that in its natural state the estuary would have supported a large open water area that would develop over several years under closed mouth conditions, followed by natural breaching after river flooding and significant sediment scour in the lower reaches and subsequent tidal exchange with the estuarine lake before the mouth closed again.
In its pristine state, the uMhlathuze Estuary was an Estuarine Lake that once connected a network of coastal marine lakes to the sea (Weerts et al. 2014). The development of a deep-water harbour at Richards Bay in the 1970s and the construction of a 4 km long berm wall divided the lake into a northern harbour section and a southern estuarine sanctuary. The uMhlathuze River to the west was canalised and diverted into the sanctuary and a new marine outlet created by dredging through the coastal berm 5 km south of the natural outlet (Weerts and Cyrus 2002). The new intertidal habitat and delta of the sanctuary was rapidly colonised by the white mangrove Avicennia marina and currently constitutes the largest mangrove habitat in South Africa (Bedin 2001;Adams et al. 2016). Invertebrate estuarine communities have become species rich from increased marine exposure, but have lost endemic species, such as the burrowing ocypodid crab, Paratylodiplax blephariskios (MacKay and Cyrus 1998). Currently both Richards Bay and uMhlathuze function as an Estuarine Bay and a Predominantly Open system, respectively (Supplementary material).
Similarly, a concrete barrage was constructed in 1978 across the estuarine connection for Lake iNhlabane to increase water storage for mining purposes. Prior to 1978, iNhlabane consisted of two interconnected lakes that had a direct connection to the sea through the iNhlabane Estuary. Raising of the water level resulted in the merging of the two lakes into a single lobed freshwater lake and the loss of typical estuarine fish and invertebrate communities (Cyrus and Wepener 1997). The downstream section of this system now functions as a Small Temporarily Closed Estuary.

Predominantly Open Estuaries
Excessive flow modifications have fundamentally changed the type and function of these systems. Freshwater abstraction and the presence of dams upstream can cause permanently open estuaries to close, as portrayed by the Uilkraals Estuary, which was once predominantly open. In 2008, it closed (seemingly permanently) for the first time and is currently experiencing successive artificial breaches and closures. In its natural state, the mouth might have closed briefly (days to weeks) during drought conditions, but current mouth closures endure for months, changing the system from functioning as a Predominantly Open Estuary to that of a Large Temporarily Closed Estuary (Table 6).
Increased mouth closure has led to abnormal brackish to fresh conditions in the Uilkraals caused by inundation levels well above the average high tide, or hypersaline conditions that result from low freshwater inputs and high evaporation. Tidal flushing has been reduced and this affects the estuary's capacity to exchange nutrients, remove accumulated salts and maintain the diversity and zonation of species-rich salt marshes (Mucina et al. 2003;Adams et al. 2010). Maintaining the system's biodiversity is important, because it falls in a biogeographic and phylogeographic transition zone between the Cool Temperate and Warm Temperate regions, with several aquatic macrophytes having phylogeographic breaks and distinct lineages within this zone (Harrison 2002;Teske et al. 2011). Closed conditions are also inhibiting for fish recruitment and estuarine nursery function.

Large Temporarily Closed Estuaries
Excessive regulated inflow can also permanently change an estuary type. Historically, the Eerste Estuary was temporarily closed and seawater intrusion created estuarine conditions up to 2.5 km from the mouth (Brown and Magoba 2009). Currently, inflow from five municipal wastewater treatment works causes the mouth to remain permanently open (Table 6). There is limited tidal influence, because seawater penetrates only 500 m into the estuary under specific mouth and river flow conditions (CSIR 2001). Fish surveys conducted before the municipal inflows commenced recorded 3-11 endemic marine and estuarine species (Clark et al. 1994). Subsequent surveys have yielded almost mono-specific catches, reduced catch rates, and the size distributions are indicative of a severe deterioration in suitable habitat for indigenous estuary-associated fish species. The invasive freshwater sharptooth catfish Clarias gariepinus is currently caught in high numbers in this once important estuarine system.
The placement of development infrastructure, such as housing, canals, railway embankments and bridges has affected the natural hydrology of numerous estuaries. Silvermine Estuary in Cape Town was once a series of large, shallow, seasonal pans and marshes that formed periodically behind a low barrier dune above the high-water mark (Brown and Magoba 2009 Table 5: Number of estuaries in each micro-system type across four biogeographical regions and classified into nine micro-system ecosystem types as a Large Temporarily Closed system, but in the 1990s, to alleviate flooding of surrounding properties, the estuary's floodplain was engineered using gabions and earth berms reducing its size, therefore changing its character to that of a Small Temporarily Closed Estuary (Table 6), with limited marine connectivity. One of the earliest maps of Table Bay, drawn by Barbier in 1786, show that the Sout-Diep system shared a common mouth. The map indicates that the Sout (Wes) Estuary was in the same location as currently, with the exception of the lower section, which used to join up with the Diep, Liesbeek and Black Rivers along the alignment of the present-day Zoarvlei, before flowing out to sea. The system is now artificially separated from the Diep, with the Diep Estuary mouth being located some three kilometres north of its historical position. Between 1930 and 2000, a series of projects were undertaken that resulted in the Sout (Wes) Estuary being diverted to the Black River and canalised in its current configuration. Extensive urbanisation of the area surrounding the Sout-Diep Estuary has largely resulted in a complete loss of estuarine function and habitat within this system.

Small Temporarily Closed Estuaries
Farther to the east, records indicate that in the early 1800s, the Small Temporarily Closed Baakens Estuary in Port Elizabeth formed a deep inlet that could be used by small sailing vessels and it was also an important recreational area. The establishment of wool-washing operations and laundries on the banks led to the water becoming polluted and unusable for recreational activities (thecasualobserver. co.za/port-elizabeth). Later, more industries were developed nearby and construction rubble was dumped into the estuary during the 1800s. This led to it becoming predominantly open and canalised along much of its length (Table 6). Over many years, almost all the estuarine natural vegetation had been removed, with only small patches of reeds remaining. Although it currently supports about nine estuary-associated fish species, the prevalence of small size classes, dominance of marine species, and lack of habitat diversity suggests that the fish fauna is severely constrained. This contrasts with the rich diversity of estuary-associated fish (30 species) found in the large permanently open Swartkops Estuary to the north (James and Harrison 2010).
The development of freeways and an industrial area close to the banks of the Small Temporarily Closed Papenkuils Estuary in Algoa Bay led to the canalisation of its lower reaches and the construction of a concrete bed ( Table  6). The estuary receives industrial and domestic effluent that pollutes its water (James and Harrison 2010). It was even known historically as Smelly Creek in the 1800s. The system currently functions as a stormwater and industrial effluent canal. Prior to these developments, typical salt marsh plants, such as Sarcocornia sp. occurred in the lower reaches, with bird life being prolific and 57 water bird species (especially flamingos) frequenting its banks (Taylor 1964). An estuary transformed into polluted stormwater canal, such as the Papenkuils Estuary, is unlikely to provide any suitable habitat for endemic estuary-associated biota (James and Harrison 2010

Mitigating for functional shifts in estuaries
Overall, where feasible, estuary management plans should ensure that natural estuarine processes in degraded estuaries are restored to ensure the overall resilience of estuarine ecosystems in South Africa. In extreme cases, some estuaries have been so extensively modified that they have completely lost functionality. Science-based management and intervention is needed to improve functionality of these modified estuaries.
Each estuary has its own issues that require remediation. The Zeekoei Estuary requires an improved stormwater infrastructure, in order to enhance its water quality. The mouth of the Heuningnes Estuary should be allowed to close, so that breaching at naturally high water levels can scour accumulated sediment.
The Uilkraals Estuary requires restoration of its baseflows to ensure the mouth remains permanently open. Recently accumulated sediment in this system might have to be removed to restore tidal flows to 2010 conditions and ensure a permanent connection to the sea. The Eerste Estuary would benefit from the establishment of an artificial wetland and decreased wastewater input to reduce excessive nutrient loading. Artificial manipulation of the Silvermine Estuary mouth and/or meandering outflow channel, by the local authorities and illegal bridge squatters, should be discontinued, because the backwater area that forms under closed mouth or meandering channel conditions is the only remaining functional estuarine area in the system. Hydrocarbons and other pollutants enter the uMhlathuze Sanctuary via the obsolete Richards Bay Harbour tidal gates should be closed off to prevent ongoing contamination of the Sanctuary area. Illegal gillnetting that is negating the uMhlathuze regionally important nursery function should also be addressed through increased compliance. The partial link between Nhlabane Lake and estuary must be established. This in turn requires continuous operation of the fishway, lowering of the barrage level, and rehabilitation of riparian areas around the lake. There are few options to improve the functioning of the canalised Sout (Wes), Baakens and Papenkuils Estuaries, apart from ensuring that pollution levels are controlled and that the water column of these systems remains oxygenated.

Conclusions
Estuary ecosystem types can serve as surrogates for ecosystem processes and enable predictions of biophysical characteristics. Understanding ecological processes and patterns associated with an estuary type facilitates an assessment of its resilience to anthropogenic pressures. This allows for extrapolation in data-limited environments.
For nearly three decades, the characterisation scheme of Whitfield (1992) served as the reference framework to type South African estuaries. This paper revises the characterisation scheme through the development of an ecosystem classification scheme that explicitly incorporates biogeographical zonation, introduces new estuary types and redefines existing types based on size. The revised classification also introduces three micro-system types previously omitted from coastal outlets on the subcontinent.
The classification scheme divided the biogeographical regions that characterise the South African coast into four major zones; the Cool Temperate (Orange to Ratel), the Warm Temperate (Heuningnes to Mendwana), the Subtropical (Mbhashe to St Lucia) and the Tropical (uMgobezeleni to Kosi), the latter being a new addition to the estuarine biogeographical provinces.
All rivers or streams with outlets to the sea were categorised broadly as estuaries and micro-systems. For estuaries, Estuarine Lakes, Estuarine Bays and Predominantly Open Estuaries were largely retained. New types, previously omitted, are Estuarine Lagoons (e.g. the groundwater-fed Langebaan system) and the Arid Predominantly Closed Estuaries that occur on the west coast (e.g. Groen and Spoeg Estauries). The numerically dominant temporarily open/closed category was subdivided into Small and Large Temporarily Closed Estuaries based on size, with a total habitat area of 15 ha serving as the threshold that separates small from large types. River mouths were renamed Fluvially Dominated Estuaries and divided into large and small size categories to reflect dissimilar catchment influences, with a total habitat area of 15 ha serving as the threshold separating types. Micro-systems were additionally divided into micro-estuaries, micro-outlets, and coastal waterfalls based on spatial features and biotic observations in the field.
Overall, South Africa's 290 estuaries were classified into 22 estuarine ecosystem categories arising from the interplay between four biogeographical zones, with nine estuary types. In addition, 202 micro-systems were also classified into nine ecosystem types, of which only the micro-estuaries (42 systems) share some functionality with estuaries. This represents a high diversity of estuary types. This is not unexpected, considering the country's diverse climatic, oceanographic and geological characteristics.
The variety of estuarine types collectively form an interconnected network of estuarine systems providing resilience to climate fluctuations and the impacts of climate change predicted for the future. Large-scale human activities, especially urbanisation and industrialisation, change ecological processes in estuaries that can permanently alter the functioning of these systems. Impacts are shown in morphological changes that affect estuary hydrodynamics and marine connectivity, chemical characteristics, such as salinity, and ultimately large-scale biodiversity loss and susceptibility to invasion by non-native species.
South African Estuarine Lakes are large and rare estuarine systems on the subcontinent. However, they are in crisis, with the majority being subjected to alterations in key ecological processes. This group of estuaries has undergone extensive infrastructure development to their functional zones, substantial flow reduction, nutrient pollution, overfishing (especially gillnetting) and manipulation of mouth areas. These combined impacts have reduced their ability to provide key ecological services, such as flood regulation, nutrient cycling and nursery habitat provision, and have compromised their value as recreation and tourism venues. To ensure their resilience to future climate-change stressors, a strategic programme is needed to restore habitat, improve water quantity and quality, reduce pressure on natural resources, and increase protection levels.
The revised classification scheme is useful because its ecosystem-level approach provides a holistic and detailed framework that integrates biogeographical factors and the extensive range of biophysical parameters evident in South African estuaries. The classification scheme forms the "blue print" for South Africa's IUCN red listing of estuarine ecosystem types, which allows for the identification of threatened ecosystem types, i.e. Critically Endangered, Endangered, or Vulnerable (Van Niekerk et al. 2019d), thus highlighting ecosystem types with an urgent necessity of management intervention and protection. Determining the condition of estuarine ecosystem types have the added advantage that it can also be used for reporting on United Nations Sustainable Development Goal 14 (Conserve and sustainably use the oceans, seas and marine resources) that specifically highlights coastal ecosystems, such as estuaries.
Ecosystem types, together with species and habitat targets, are used in setting conservation planning targets to ensure that all life supporting abiotic and biotic processes are captured in a representative protected areas network . Given that the identified estuary types characterise physical and biotic processes, they can also be used as proxies for predicting sensitivity to anthropogenic pressures, such as flow reduction and increased nutrient loading, in environmental flow assessment in data-poor environments. Aquatic ecosystem typing is one of the fundamental datasets for extrapolating freshwater flow requirements across a region in low confidence assessments (van Niekerk et al. 2015).
Although this study focused on developing a data-driven ecosystem classification system, key datasets were found to be lacking that could facilitate a more detailed assessment. Critical parameters that would have allowed for a more nested or hierarchal classification include information on seasonal salinity regimes, mouth states, water clarity, estuary topography and bathymetry, and sediment structure. Accurate records/observations of the duration of the closed mouth condition are especially critical for future classification updates. In addition, very little information was available on the invertebrate community for most of South Africa estuaries. Overall, data on biological responses are more than 30 years old (e.g. national bird counts date from the 1980s and national fish surveys date from the 1990s). While new data are being collected on some of the larger systems, very little information is currently being gathered on the numerous smaller estuaries, especially those in remote areas along the coast. Without a major investment in baseline information from numerous poorly studied South African estuaries, it will be difficult to refine and possibly expand this new classification system in the future.
The classification of estuaries is not an exact science, because some systems demonstrate biophysical features that fit criteria of more than one type. The uMthavuna is a Predominantly Open Estuary (mouth open >90% of the time); however, on most recent satellite imagery the mouth is highly constricted and more representative of a Large Temporarily Closed system in a semi-closed or closed condition. Similarly, some Subtropical Predominantly Open Estuaries are fed by sediment-rich catchments and they therefore meet some of the criteria of a Large Fluvially Dominated type, e.g. the uMzimkhulu and uMkhomazi Estuaries. Classification schemes should therefore not be divorced from the context within which they are applied, e.g. an evaluation of important systems contributing to coastal sediment supply cannot only focus on Large Fluvially Dominated systems, but should also include sediment-rich Predominantly Open Estuaries. Although classification schemes strive to provide structure and guidance on estuarine biophysical processes and features, estuary classification anomalies are to be expected along such a geological and climatic diverse coastline.