Mangrove and sand cay dynamics on Australian and Indonesian low wooded islands: A 45 year comparison of changes from remote sensing

Highlights

• We compare sand cay and mangroves changes on Australian and Indonesian reef islands.

• Changes are observed for the GBR, Australia and Spermonde archipelago, Indonesia.

• Spermonde and GBR sites have distinctive the sedimentary and anthropogenic dynamics.

• Changes on reef islands depend on local environmental and anthropogenic influences.

Abstract

Changes to coral reef landscapes are driven by regional processes that are unique to particular localities, yet much of our global knowledge about how landscape changes manifest in coral reef environments is generalised from work undertaken on the Great Barrier Reef. We compare observations of 45 years of change on sand cays and mangroves associated with low wooded islands in Australia and Indonesia. We draw on field observations from ground referencing campaigns, alongside remote sensing technology, including satellite images and unmanned aerial vehicle campaigns. Four low wooded island sites are compared: two in the Great Barrier Reef (GBR), Australia (Nymph Island and Two Isles) and two in the Spermonde Archipelago, Indonesia (Sabangko and Tanakeke Island). The Spermonde and GBR sites can be distinguished in relation to the process regimes that entrain, distribute and deposit sediments on the reef surface thereby providing a substrate for further mangrove colonisation, particularly the presence or absence of cyclones as a key determinant of sediment transport. The influence of human populations inhabiting these sites is also an important control on their geomorphology. In the Spermonde Archipelago, local communities have altered sand cays through the development of infrastructure and converted mangroves to shrimp farms, while sand cays and mangroves have remained largely unaltered by humans on the GBR. This comparative evaluation of changes to sand cays and mangrove forest across low wooded islands emphasises the importance of considering changes within the context of their local geographic setting, inclusive of natural environmental and anthropogenic drivers of change.

Introduction

Globally, reef islands are extremely significant land areas on which small island developing states have built substantial infrastructure to support populations in the Atlantic, Pacific and Indian Oceans. The upper surfaces of coral reef platforms are highly dynamic coastal environments in which sand cays migrate, mangrove forests extend and contract, and the terminal spits of sand cay beaches and shingle ridges move substantially. A better understanding of reef island responses to environmental and anthropogenic influences is critically important for people whose homes and livelihoods depend on the goods and services they provide.

It is widely thought that environmental changes such as sea level-rise will cause reef islands to erode or even disappear by the end of the century. This idea is both supported and refuted by recent shoreline changes studies in the Pacific and Indian oceans (McLean and Kench, 2015). The stability of reef islands relies on a locally-defined relationship between sea level and the hydrodynamic and carbonate production processes that produce and transport sand in and around reef flats. These processes fundamentally control sand cay formation and sedimentation, which facilitates mangrove propagule establishment (Smithers and Hopley, 2011). All of these factors are geographically structured in the sense that sea level histories vary regionally, along with contemporary environmental processes, including cyclone frequency and intensity, the hydrodynamic climate (tidal range and wave power) and the ways in which local human populations have inhabited and interacted with the reef island landscapes. Thus, an understanding of recent changes on reef top landforms at both the global and regional scale calls for an awareness of the geographic distribution of the factors of importance to these landforms, how these are realised locally, and also the limits to which insights gained in one geographic area can be reliably applied elsewhere.

Found throughout the Caribbean, Pacific and Indian Oceans, low wooded islands are a distinct type of island that forms on reef platforms, with a sand cay on the leeward reef flat and mangroves on the exposed reef flat (Hopley et al., 2007; Kench, 2011). We focus on low wooded islands because the interrelations between the leeward sand cay and reef flat mangrove depend on their local environmental setting.

Historically, much of the work done on the geomorphic behaviour of low wooded islands has been carried out on the Great Barrier Reef (GBR) as a result of two Royal Society Expeditions in 1929 (Steers, 1929) and 1973 (Stoddart, 1978a), as summarised in Hamylton et al. (2019). Stratigraphic evidence and sediment dating at Bewick and Pipon Islands have identified a ‘reef platform process window’ of coral reef development and formation during which a critical water depth over the reef platform enhanced sediment production and transport to form islands within a rapid timeframe, some 5000-4000 yrs B.P. (Kench et al., 2012; Perry et al., 2017).

On the GBR, it has been suggested that low wooded islands form on top of planar reefs, at the end of a geomorphic evolutionary sequence in which underlying reef platforms grow upward to sea level from a Pleistocene basement, then expand laterally while their upper surface undergoes sediment infill. This is followed by development of a leeward sand cay, reef flat mangroves and shingle ramparts, which commonly form across the windward shoreline (Hopley, 1982; Steers and Kemp, 1937). Carbonate sediment deposition initiates a set of rapid geomorphic changes that can replace bare reef flats with a complex array of landforms over decadal timescales, invariably filling up the entire reef flat over longer periods (Stoddart et al., 1978). Once this point at the end of the proposed evolutionary sequence is reached, minor changes to low wooded islands reflect adjustments to localised fluctuations in hydrodynamic and climatic conditions, within a broader equilibrium with local environmental conditions (Spender, 1930).

It is difficult to propose a unifying model of low wooded island evolution that incorporates the global range of environmental conditions observed in different regional sites. The aforementioned equilibrium described on the GBR is contingent on the islands being unmodified. Yet in many countries, low wooded islands are heavily populated and adjustments to the reef top landforms also reflect the activities of communities inhabiting and deriving their livelihoods from islands.

We adopt a comparative approach to evaluate geomorphic changes to low wooded islands from different geographic settings that are subject to unique localised anthropogenic and environmental processes. Comparative approaches have a long standing place in geographic enquiry that draws useful explanatory power and insights from regional case studies that enable islands to be systematically categorised and compared within their geographic setting. Sites in both the Australian GBR and the Indonesian Spermonde Archipelago represent distinct environmental settings that fall under different anthropogenic regimes. Although both sites support low wooded islands within inshore, turbid environments, the differences and similarities in their emergent landform behaviours may yield new insights about low wooded island geomorphic behaviour that apply across a broader geographic range than has hitherto been the case.

The need for a better understanding of both the environmental and anthropogenic controls that govern the ongoing morphodynamics of islands at seasonal to decadal timeframes has been emphasised for both the GBR (Hamylton et al., 2019) and the Spermonde Archipelago (Kench and Mann, 2017). Early comparisons between these island groups arose from the Great Barrier Reef Expedition (1928–29) and the Snellius Expedition to the eastern part of the Netherlands East Indies (1929–1930) (Kuenen, 1933; Spender, 1930), which noted the similarity in these inshore islands, concluding that they could both be classified broadly as low wooded islands. Since then, the rapid development of housing infrastructure on sand cays and mangrove aquaculture across reef flats in the Spermonde Archipelago has resulted in markedly different coastal changes in these different settings. Here, we update those comparisons using satellite and unmanned aerial vehicle (UAV) remote sensing approaches to evaluate 45 years of sand cay and mangrove dynamics as a basis for exploring and comparing the processes underpinning both geomorphic and anthropogenic changes in these distinct environments. The following comparisons are drawn for two low wooded islands from each site: Nymph Island and Two Isles on the GBR and Sabangko Island and Tanakeke Island in the Spermonde Archipelago (Fig. 1):

• Changes to the position of shorelines around sand cays, including seaward accretion and inland erosion. For the Indonesian sites, we track the development of infrastructure across sand cays.

• Changes to the boundaries of the reef top mangrove forest. For the Indonesian sites, we track the conversion of mangrove forest to aquaculture ponds across the reef flat.

• Three dimensional characteristics of the sand cay shorelines and mangroves are evaluated for both of the Australian sites and one of the Indonesian sites (Sabangko) from available high resolution digital elevation models (DEM).

The GBR is the largest reef province in the world. It stretches over 2000 km of latitude, with >3000 individual reef patches extending from the mainland coast some 250 km offshore to the continental shelf edge. These reef patches provide a foundation for over 1000 reef islands, including 350 coral cays (Smithers and Hopley, 2011). These include 47 low wooded islands, on the inner shelf of the northern GBR.

The GBR is situated in the centre of the Indo-Australian plate, imparting tectonic stability to this region. The last glacial transgression produced varying relative sea level histories along and across the continental shelf of the GBR, resulting in a variable time range over which islands have accumulated (Hopley et al., 2007). The reefs in this region grew up from the shallow continental shelf (<50 m water depth) to current sea level over the mid to late Holocene (i.e. the last 6.5 k years), providing an intertidal platform for upper and lower terraces of island deposition dating back some 3000–4000 years, as indicated by radiocarbon dates of island sediments on the northern GBR (Hopley et al., 2007; McLean and Stoddart, 1978).

Semidiurnal tides flow on and off the continental shelf, resulting in a considerable movement of water along an east–west axis, with a tidal amplitude ranging from 2.5 in the northern and southern extents, and reaching a maximum of 6.1 m in the southern region of Broad Sound (Wolanski, 2018). Tropical cyclones form between Cape York and Brisbane, with a pronounced late-summer peak when sea conditions are hottest between November and April (Harmelin-Vivien, 1994; Short and Woodroffe, 2009). The eastern boundary of the GBR intercepts swell waves from the Pacific Ocean (Gallop et al., 2014). Most islands are subject to smaller trade wind waves generated within the GBR lagoon, with inner reefs experiencing significant wave heights of up to 0.8 m for locally wind-generated waves (Larcombe et al., 2001). WaveWatch III (WW3) Global Wave Model data (Tolman, 2009) for the period 2013–2019 indicate yearly average significant wave heights of 0.51 m for Nymph Island and 0.63 m for Two Isles, with prevailing winds come from a southeasterly direction for the majority (85%) of the year, except during summer months (December to February) when the northwest monsoon dominates. This monsoon is characterized by heavy rainfall, occasional cyclones and a dominant southeasterly swell that transports carbonate sediments length-wise across many reef platforms, (i.e. in a northwesterly direction) (Hamylton et al., 2019). Carbonate sediments are derived from the periphery of the shallow outer reef slope, particularly on the windward side. High energy events including storms and cyclones generate and remobilize sediment around the reef flats of several low wooded islands in the northern portion of the GBR, often forming shingle ridges comprising coarse coral fragments, around windward reef margins (Moorhouse, 1936; Perry et al., 2014). During the passage of cyclones, significant wave heights rise to around 1.9–2.2 m, as indicated by WW3 data for cyclone Cyclone Ita, which passed close to both Nymph and Two Isles in 2014.

The Makassar Strait opened up in Central Indonesia some 20,000 years ago with the separation of Borneo from Sulawesi (Imran et al., 2013). On the eastern coast of the Strait, the 30–50 km wide Spermonde Shelf sits along the shore of south-western Sulawesi. The Spermonde Archipelago consists of approximately 54 reef islands that have formed over coral reef patches across this shelf. The nearest major city is Makassar on an adjacent alluvial coastal plain, which has a population of 2.75 million (Fig. 1).

Relative to the GBR, the Spermonde Shelf is in a tectonically active area, situated close to the confluence boundaries of the Philippine Sea plate, Eurasian plate and the Indo-Australian plates. This is just outside the ‘Pacific Ring of Fire’, a region subject to frequent earthquakes and volcanoes that cause uplift and tsunamis.

Approximately 137 coral reefs are supported by the Spermonde Shelf, which can be subdivided into the inner, middle and outer zones with localised environments characterised by distinct water depths, hydrodynamics and fluvial influence from the adjacent coastline (Nurdin, 2019). Two major rivers discharge onto the Spermonde shelf, the Jeneberang which runs along the southern margin of Makassar City and the Maros River, approximately 20 km to the north. These deliver clastic sediments to inshore reefs and islands, (<4 km from the coast), reducing the clarity and elevating levels of nutrients of coastal waters.

The ages of fossil microatolls in the Makassar Strait indicate that the reef platform reached current sea level here some ~6500–6000 years ago (Mann et al., 2016). This is in agreement with Umbgrove's view that the present coral reefs on the Spermonde shelf must have originated at the end of the Pleistocene, rising from a depth of 30 fathoms (approx. 50 m) to form the many submarine reef platforms, patch reefs and cays in variable stages of development that are seen today (Umbgrove, 1928).

Population growth across the Spermonde Islands has increased substantially since the 1940s, with extensive housing development on 50 of the 54 islands (Fig. 2). Interviews with community members on the most populated island, Barrang Lompo, which is located 12 km west of Makassar city and, according to the Central Indonesian Bureau of Statistics housed some 3696 people in 2018, indicate that it was first populated 74 years ago (Nurdin, 2019). Communities living on the islands depend on coastal resources for their livelihoods. Key sources of income include coral reef fisheries, seaweed farming, aquaculture (prawn farming) and the collection of crabs, squid, octopus and shellfish (Fujii, 2019; Glaser et al., 2015). Reef flats are commonly covered with live coral and seagrass, although dynamite fishing and the run-off of pollution from Makassar have led to regional declines in coral health (Edinger et al., 1998; Fujii, 2017; Nurdin et al., 2015).

The movement of surface water currents across the Spermonde platform depends on the velocity of monsoon-driven winds and tides (Jalil, 2011). Tides are semidiurnal within the Makassar Strait with a small range, typically between 0.2 and 0.3 m. Sulawesi sits at 5° south, around 500 km to the north of the tropical cyclone belt. As a result, shingle ridges are less well developed on many of the patch reefs in the Spermonde Archipelago, reaching modest proportions in the places they do exist, where ‘no part is exposed at normal low water’ (Verstappen, 1954).

Monsoons have a strong, seasonal influence on regional winds and waves. Another key influence on regional hydrodynamics is the large volume of water known as the Indonesian Throughflow (ITF) passes through the Strait, which provides a major conduit of largely southward flowing water between The Pacific Ocean and the Indian Ocean (Jompa, 1996). The monsoonal regime drives winds to the southeast during the southern hemisphere summer (December–February), with corresponding currents from the north that flow down from the northern Pacific. During the winter (June–August), winds blow toward the northwest with surface currents coming up from the Java Sea to the south, flowing against the ITF.

In general, the northern Spermonde Archipelago falls within a region of low to moderate wave energy (mean significant wave height 0.7 m), with the Sulawesi mainland moderating the influence of easterly wind patterns (Umbgrove, 1929). Tanakeke occupies a more exposed position, approximately 10 km off the far southwestern tip of Sulawesi with wave heights commonly reaching 2 m. Further north in the centre of the Spermonde Shelf, Sabangko lies some 4 km offshore and experiences a relatively consistent annual average significant wave height of 0.2 m (Tolman, 2009).