298

IPCC90, the overall judgment is that confidence in the predicted patterns of climate change at spatial scales relevant to coastal zones and small islands must still be regarded as low (Gates et al., 1992, 1990; Mitchell et al., 1990; McGregor and Walsh, 1993). Nonetheless, there are some broad regional results that are consistently produced by equilibrium and transient GCM experiments that have relevance to the coastal zone. These results include (Mitchell et al., 1990; Gates et al., 1992):

The mean surface air temperature increases more over land than over oceans (by about a factor of two), and at higher latitudes during winter. The climate of the coastal zone tends to be moderated by ocean temperature and will thus be strongly influenced by sea-surface temperature changes. Northern Hemisphere warming is greater than Southern Hemisphere warming because of the larger extent of land.

In general, global models show increases in precipitation throughout the year in the high latitudes and during the winter in mid-latitudes. Most models show some increase in Asian monsoon rainfall.

At scales relevant to coastal zones and small islands, predictions of changes in other climate elements, such as windiness, storminess or radiation, cannot yet be considered reliable.

Although changes in sea-surface temperatures (SSTs) are projected to be less than those on land, they are not necessarily less significant. As Edwards (1995) has pointed out, a 2°C change in SST in the tropical and subtropical oceans is considered to be anomalous, but is on the order of temperature changes associated with strong El Niño/Southern Oscillation (ENSO) events. In comparison, the projected change in mean sea-surface temperature for these regions is on the order of 1-2°C by the year 2100. Thus, by 2100, SSTs that are now considered anomalous could well be normal occurrences. Such warming would be unprecedented in the recent geological past.

Tropical cyclones (also known as hurricanes or typhoons, depending on region) affect vast coastal areas in tropical and subtropical countries. Tropical cyclones and associated storm surges can cause enormous loss of life and have devastating impacts on coastal ecosystems and morphology. It is therefore of critical importance to know how the frequency, magnitude, and areal occurrence of such storms will change in a warmer world, if at all. Unfortunately, the evidence from theoretical and numerical models and from observational data is, as yet, inconclusive. While current GCMs provide some indication of possible tropical cyclone formation, identifying "tropical disturbances" (e.g., see Broccoli and Manabe, 1990; Haarsma et al., 1993), they cannot explicitly model such storms at the present grid-scale resolution (Mitchell et al., 1990). However, recent work has been moving in this direction (e.g., Bengtsson et al., 1994). Alternatively, theoretical storm models have been used to examine the maximum storm intensity in relation to SSTs (Emanuel, 1987) but cannot easily be extended to address questions of regional changes in tropical storm intensity or frequency under conditions of climate

Coastal Zones and Small Islands

At present, there is no evidence of any systematic shift in storm tracks. The tracks are governed by the location of cyclogenesis and prevailing meteorological conditions; thus far, there is no evidence of shift in the preferred locations of cyclogenesis. Empirical studies have found correlations between ENSO and the regional patterns of tropical cyclone activity (as well as the Southeast Asian monsoon, Atlantic hurricanes, Pacific precipitation patterns, and other phenomena) (Nicholls, 1984; Evans and Allan, 1992; see also Chapter 3. Observed Climate Variability and Change, of the IPCC Working Group I volume). Progress is being made on model simulations of the present features of ENSO-like events (e.g., Philander et al., 1992). which could lead to a predictive capability for the future. Recent coupled-model simulations under enhanced CO2 show a tendency toward ENSO-like patterns in the Pacific, with accompanying temperature and precipitation variability (see Chapter 6, Climate Models-Projections of Future Climate, of the IPCC Working Group I volume), but realistic simulations of ENSO are not yet possible. The behavior of ENSO is critical in understanding the future coastal effects of both climate change and sea-level rise in the Pacific region and elsewhere (Pittock and Flather, 1993; Pittock, 1993). In short, it is not yet possible to say whether either the intensity or frequency of tropical cyclones (or ENSO) would increase or the areas of Occurrence would shift in a warmer world.

Despite the often repeated assertion that climate variability could increase in a warmer world, there is little evidence from climate models to support this notion (Gates et al., 1992). However, Working Group I has identified at least one exception that has potentially large implications for coastal areas: Various GCMs consistently predict a higher frequency of convective precipitation in mid- to high-latitude regions of the world. This anticipated change may imply more intense local rainfall, with a decrease in the return period of extreme rainfall events (e.g., Gordon et al., 1992). This could interact with sea-level rise to further increase the likelihood of flooding in low-lying coastal areas (Titus et al., 1987; Nicholls et al., 1995).

Without any change in variability, however, a change in mean value still implies a change in the frequency of extreme events. Because such events by definition are at the tails of the proba bility distribution, the change in return periods can be quite large relative to the change in mean value. In the absence of information about changes in variability, this simple concept has been employed to create scenarios of changes in extremes-for example, for temperature in Britain (Warrick and Barrow, 1991), rainfall in Australia (Pittock et al., 1991), and storm surges in the United States (Stakhiv et al., 1991). As has been pointed out at the beginning of this section, in most cases it is the combination of climate extremes-temperature. precipitation, winds, sea levels-and how they are affected by longer-term changes in climatic means that are especially important for considering the future effects of global warming on coastal zones and small islands. Yet little understanding of the possible interaction of different aspects of climate change

It was recognized then--and highlighted again in the IPCC 1992 supplement (Tsyban et al., 1992) and at the meeting The Rising Challenge of the Sea (IPCC CZMS, 1992; O'Callahan, 1994) that such effects would not be uniform around the world and that certain coastal environments would be especially at risk. These included tidal deltas and low-lying coastal plains, sandy beaches and barrier islands, coastal wetlands, estuaries and lagoons, mangroves, and coral reefs. Small islands became a major focus of concern because some of the more extreme predictions foreshadowed that low atoll and reef islands would completely disappear or become uninhabitable, with the total displacement of populations of several small island nations (Roy and Connell, 1991).

Many of the early studies on the effects of climate change emphasized sea-level rise and were based on a simple inundation model that vertically shifted the land-sea boundary landward by the amount of the projected global rise. However, it has become increasingly clear from recent studies that the geomorphological and ecological responses to a rising sea level will be complex and will also reflect a large number of other factors, including other aspects of climate change. No longer can effects be defined simply in terms of inundation of the sea upon the land, nor by just shifting the land-sea contour by an amount corresponding to the projected vertical increase in global sea level. Biogeophysical effects will vary greatly in different coastal zones around the world because coastal landforms and ecosystems are dynamic and both respond to and modify the variety of external and internal processes that affect them. Effects will depend not only on the local pattern of sea-level rise and climate change (as shown in Section 9.3) but also on the nature of the local coastal environment and on the human, ecological, and physical responsiveness of the particular coastal system being considered (J.R. French et al., 1995).

Since the IPCC 1990 assessment, considerable progress has been made in understanding the effects of sea-level rise and climate change on coastal geomorphological and ecological systems. Studies have shifted from the use of simple, monothematic approaches to more complex yet pragmatic

In all three cases, emphasis has been on sea-level rise, with little consideration of other climate-change aspects, although sometimes increased seawater temperatures and storminess have been included. Invariably, global sea-level rise scenarios have been applied, irrespective of their appropriateness at the local or regional level. Further, several authors (e.g., Bird, 1993a; J.R. French et al., 1995) have argued that it is not always appropriate to employ Holocene stratigraphical reconstructions as analogues for the future behavior of coastal systems, primarily because of the modern complications of human impacts that may now have an overriding effect on geomorphological and ecological responses.

In spite of the increased research effort, there is still no generally accepted global typology of coastal types relating to the potential effects of sea-level rise and climate change. There have been some attempts based on the resistance of the coast to environmental forces (e.g., Van der Weide, 1993), some based on both natural and socioeconomic features and processes (Pernetta and Milliman, 1995), and some through the development of a coastal vulnerability index that captures the different characteristics of a coastal region (e.g., Gornitz, 1991). The development of such a typology is clearly an area for substantial international research in the future. Moreover, the emphasis of recent studies on coastal types has been quite uneven. For instance, there has been little research on the potential effects of sea-level rise and climate change on high-latitude coasts, bold coasts, rocky shores, coastal cliffs, coarse clastic coasts, gravel barriers, coastal sand dunes, and seagrass beds. This lack of emphasis, however, does not necessarily imply that anticipated effects on these coastal types are less serious. Studies that show a strong correlation between sea-level rise and erosion of, for example, coastal cliffs, gravel barriers, and sand dunes include Griggs and Trenhaile (1994), Carter and Orford (1993), and Van der Meulen et al. (1991), respectively.

There are several comprehensive reviews on the biogeophysical effects of climate and sea-level change on coastal environments (e.g., Bird, 1993b; Oude Essink et al., 1993; Wolff et al., 1993). In addition to these general reviews there is a series of regional summaries covering a large area of the worldincluding the Mediterranean (Jeftic et al., 1992). European coastal lowlands (Tooley and Jelgersma, 1992), Southeast Asia (Bird, 1993a), the South Pacific (Hay and Kaluwin, 1993), wider Caribbean (Maul, 1993), the Western Hemisphere

57-716 99-18

Open coasts, primarily made up of unconsolidated sands and gravels and exposed to wind and wave action, are common on all inhabited continents and islands of all sizes. About 20% of the world's coast is sandy and backed by beach ridges, dunes, or other sandy deposits. International studies reported by Bird (1985, 1993b) indicate that over the last 100 years about 70% of the world's sandy shorelines have been retreating; about 20-30% have been stable and less than 10% advancing. He has listed at least twenty possible reasons for the prevalence of erosion and has indicated that sea-level rise is only one possibility. Although Stive et al. (1990), Leatherman (1991), and others have recognized a causal relationship between erosion and sealevel rise, many attempts to correlate accelerated coastal erosion with global sea-level rise over the last 100 years have not been convincing because of the difficulties in excluding other factors, including human impacts. Analyses of erosional trends on sandy shorelines over the past several decades indicate a predominance of local rather than common explanations—suggesting that, if sea-level rise has been a contributor, its contribution may have been masked by other mechanisms.

Two other approaches have been used to gauge the effect of sea-level rise on sedimentary coasts. First, models have been used to predict beach-profile changes that will result from a rise in water level. Model studies have been reviewed by international expert committees such as the Scientific Committee on Oceanic Research (SCOR, 1991), as well as by individuals (e.g., Healy, 1991; Leatherman, 1991). The best-known model is that of Bruun (1962), who formulated a two-dimensional relationship between rising sea level and the rate of shoreline recession based on the concept of profile equilibrium, which has been the subject of much evaluation (e.g., Dubois, 1992). SCOR (1991) has noted that testing and application of the models for beach response to a long-term rise in sea level have been hampered by significant lag times of beach changesamounting to months or years-and the importance of other elements of the sediment budget that produce shoreline erosion or accretion irrespective of any sea-level rise. Profile changes assumed by the models have been reasonably well-verified by laboratory and field studies, but the predictive equations are found to yield poor results when the effects of profile lag times and complete sediment budgets are not included in the analysis. One solution to these uncertainties is to determine a range of beach-recession scenarios rather than a single estimate

Coastal Zones and Small Islands

although SCOR (1991) has concluded that the status of models for the beach response to elevated water levels is far from satisfactory: predictions of the associated shoreline recession rates yield uncertain results; and there is clear need for substantial research efforts (field and laboratory) in this area. A new generation of shoreface-profile evolution models is presently being developed (e.g., Stive and De Vriend, 1995).

Second, morphostratigraphic studies, particularly of sandy barriers, have been undertaken, although frequently these studies predate the recent interest in attempting to predict future coastal response to climate change and sea-level rise. Nevertheless, sandy-barrier responses to rises in sea level in the Holocene can be used as historical analogues. Although transgressive sedimentary sequences, where coastal barriers migrate landward as a result of shoreface erosion and washover, are widespread in North America, Europe, and Australia, other responses to sea-level rise include in situ growth (the stationary barrier) and even seaward advance (the regressive sequence). As with other approaches, field-based evolutionary morphostratigraphic models do not yield a consistent response to sea-level rise. Rate of sediment supply and coastal configuration are just two of the other factors that influence how sandy shorelines will respond. In addition to field-based studies, some indication of the complex way that sand barriers have responded to post-glacial sea-level rise has been shown through computer-simulation techniques (Cowell and Thom, 1994; Roy et al., 1994).

Deltas form where terrigenous sediment brought down to the coast by rivers accumulates more rapidly than can be removed by waves, tides, and currents. Although there is a wide spectrum of delta types around the world, all are the result of the interaction between fluvial and marine processes. Since ancient times, deltas have been of fundamental importance to civilizations due to the presence of highly productive agricultural lands, fisheries, and human settlement. Many modern delta regions, with their dense populations and intensive economic activities, are now in crisis because of past management practices such as dam, dyke, and canal construction and habitat destruction, which have led to problems such as enhanced subsidence and reduced accretion, salinity intrusion, water quality deterioration, and decreased biological production (Day et al., 1993; Boesch et al., 1994).

Deltaic coasts are particularly susceptible to any acceleration in the rate of sea-level rise (as well as storm frequency or intensity). As Baumann et al. (1984) have recognized, delta survival

Coastal Zones and Small Islands

is a battle of sedimentation versus coastal submergence. Most deltas are subsiding under the weight of accumulating sediment, a process that often is enhanced by artificial groundwater withdrawal. Any global sea-level rise will exacerbate existing problems of local submergence. Bird (1993b) has argued that a rising sea level will have two major effects on low-lying deltaic areas: First, it is likely to cause extensive submergence, especially where there is little prospect of compensating sediment accretion. Second, progradation of most deltaic coastlines will be curbed, with erosion becoming more extensive and more rapid.

Similar conclusions have come from a host of case studies around the world, including those reported from Europe and the Mediterranean in Jeftic et al. (1992), Tooley and Jelgersma (1992), Poulos et al. (1994), and Woodroffe (1994); from the Americas in Day et al. (1993, 1994); and from Southeast Asia in McLean and Mimura (1993). While there appears to be general agreement among all of the studies on the implications of reduced sediment discharge, subsidence, and rising sea level, there have been few attempts to determine the relative vulnerability of deltaic regions or to model the effects of sea-level rise on deltas. Exceptions to the former include Ren's (1994) study of the Chinese coast, in which six variables (relief, land subsidence, shoreline displacement, storm surge, tidal range, and coastal defenses) have been used to evaluate risk classes of eight vulnerable areas. Exceptions to the latter include the conceptual model of general deltaic functioning developed by Day et al. (1994) and their two-state variable model, which simulates height in sea level and land elevation over time as a function of varying rates of sea-level rise, subsidence, and vertical accretion. Model results of the "date of immersion" (i.e., when sea level equals land elevation) have been produced for several sites in the Mississippi, Camargue, and Ebro deltas. Intradelta variations have been highlighted in the model results. In natural situations, such variations commonly result from variations in subsidence and/or changes in active and passive distributary positions across deltas, as demonstrated for the Nile (Stanley and Warne, 1993) and Rhine-Meuse (Tornqvist, 1993) deltas, respectively.

Studies on the physical response of tidal rivers and estuaries to predicted sea-level rise have covered two main areas-geomorphic changes and saltwater penetration-although the American Society of Civil Engineers (ASCE) Task Committee (1992) has indicated that several hydraulic processes such as tidal range, prism and currents, and sedimentation would also be modified. Bird (1993b) has suggested that estuaries will tend to widen and deepen. This may enhance their role as sediment sinks, causing greater erosion of the neighboring open coast (Stive et al., 1990). However, Pethick (1993) has shown that along the southeast coast of Britain, where relative sealevel rise is already 4-5 mm/yr due to local subsidence, estuarine channels are becoming wider and shallower by local redistribution of sediment as the intertidal profile shifts both upward and shoreward. In some areas, these effects may be offset by increased catchment runoff, greater soil erosion, and increased sediment yield as a result of climate changes.

301

However, as with deltas, critical factors will be relative sealevel change, including local subsidence (e.g., Belperio, 1993), and sediment availability (e.g., Chappell, 1990; Parkinson et al., 1994). In macrotidal estuaries in Northern Australia, channel widening initiated by rising sea level will contribute sediment to the adjacent estuarine plains, which may offset the effect of flooding and lead to steady vertical accretion. One consequence of this would be to endanger backwater swamps and freshwater ecosystems on the estuarine plains (Chappell and Woodroffe, 1994).

The effects of sea-level rise on saltwater penetration in rivers and estuaries have recently been reviewed by Oude Essink et al. (1993) and Van Dam (1993), who have suggested that saline water will gradually extend further upstream in the future. More serious is the accelerated effect of saline water intruding into groundwater aquifers in deltaic regions and coastal plains. In these areas, the effect of sea-level rise can be exacerbated by the withdrawal of freshwater, which may result in either subsidence and/or replacement by seawater. Subsidence and landward migration of saltwater are already serious problems in many coastal deltaic areas around the world. Two examples are Myanmar (Aung, 1993) and China (Han et al., 1995b).

9.4.3. Coral Atolls and Reef Islands

Coral atolls and reef islands appear especially susceptible to climate change and sea-level rise. Based on the sea-level rise scenarios of the 1980s and the application of simple models, Pernetta (1988) developed an index of island susceptibility for the South Pacific region and concluded that the most susceptible nations included those "composed entirely of atolls and raised coral islands, which will be devastated if projected rises occur," and consequently "such states may cease to contain habitable islands." Three related effects were envisaged: erosion of the coastline, inundation and increased flooding of lowlying areas, and seawater intrusion into the groundwater lens, which would cause reductions in island size, freeboard, and water quality, respectively.

Since that time, a series of vulnerability assessments of atolls and reef islands have been carried out. Studies include the atoll states and territories of Tuvalu, Kiribati, Tokelau, and the Marshall Islands in the Pacific and the Maldives and Cocos (Keeling) Islands in the Indian Ocean (Aalbersberg and Hay, 1993; Woodroffe and McLean, 1992; McLean and D'Aubert, 1993; Holthus et al., 1992; Connell and Maata, 1992; Pernetta, 1992; McLean and Woodroffe, 1993). Generally, these studies have documented the likelihood of more complex and variable responses than initially suggested, recognizing that the balance between reef growth, island accumulation or destruction, and sea-level rise will be locally important. Differences in response can be further expected between islands within and beyond storm belts, between those composed primarily of sand and those of coral rubble, and between those that are or are not anchored to emergent rock platforms. The presence or absence of natural physical shore-protection structures in the form of

302

beachrock or conglomerate outcrops and biotic protection in the form of mangrove or other strand vegetation will also result in different responses between islands.

It is not clear to what extent reef islands will erode or whether sediment from the adjacent reef or lagoon will contribute to the continued growth of islands. McLean and Woodroffe (1993) have envisaged at least three possible responses in the face of sea-level rise: the Bruun response, the equilibrium response, and continued growth, which would result in shoreline erosion, redistribution of sediment, and shoreline accretion, respectively. Each of these processes can be observed on many reef islands today, as well as in the stratigraphic record, suggesting that the factors identified above are significant determinants of island stability. Moreover, as Spencer (1995) has pointed out, coral-island responses to future sea-level rise will vary as a result of constraints on the development of modern reefs and the varying inherited topographies upon which future sea-level will be superimposed.

On small islands, the freshwater lens is an important resource and often is the primary source of potable water on atolls. Recent studies suggest that the first approximation of the response of the freshwater lens to sea-level rise (the GhybenHerzberg principle) is not appropriate on small coral islands. The layered-aquifer model-which, among other things, considers geological structure and distinguishes between Pleistocene and Holocene stratigraphic units-is considered more appropriate for assessing freshwater inventories on such islands. If recharge and island width remain constant or expand, freshwater lenses may actually increase in size with a rise in sea level because of the larger volume of freshwater that can be stored in the less-permeable upper (Holocene) aquifer (Buddemeier and Oberdorfer, 1990). On the other hand, if recharge or island width are reduced, a diminution in both freshwater quantity and quality can be expected. In many places, increasing demand and recharge contamination are likely to be more serious issues than freshwater inventory per se.

Although recent reviews on coral islands have emphasized their variability and resilience (e.g., Hopley, 1993; McLean and Woodroffe, 1993), such islands remain among the most sensitive environments to long-term climate change and sealevel rise, especially where these effects are superimposed on destructive short-term events such as hurricanes, damaging human activities, and declining environmental quality. In spite of a more optimistic outlook in recent years, Wilkinson and Buddemeier (1994) have maintained that coral-reef islands may be rendered uninhabitable by climate change, especially sea-level rise, and that will necessitate relocation of any remaining human populations (see also Section 9.5).

9.4.4. Coastal Wetlands

Coastal wetlands are frequently associated with deltas, tidal rivers, estuaries, and sheltered bays. Geomorphic and hydrologic changes resulting from sea-level rise will have important

Coastal Zones and Small Islands

effects on these biological communities, as well as on unvegetated tidal flats. The survival of the latter is dependent very much on sediment supply from adjacent river catchmentswhich, if not provided, will result in substantial loss of such areas. Although Woodroffe (1993) has commented that research on coastal wetlands has concentrated upon reconstructing their development under conditions of sea-level rise during the Holocene, there also have been assessments of contemporary trends and processes and simulation modeling of environmental changes.

Historical studies of temperate salt marshes include those of Allen (1991) and Reed (1990), whereas Pethick (1993) and French (1993) have used current trends and numerical simulation, respectively. Pethick (1993) has shown that salt marshes in southeast England appear to be migrating inland along the estuary but that the natural changes are interrupted by the presence of flood embankments. The result is that loss of the seaward boundaries of these wetlands will continue without compensating landward migration-a process known as coastal squeeze in the United Kingdom. Wolff et al. (1993) have concluded that salt marshes have the ability to respond quickly to sea-level rise as long as sedimentation and internal biomass production processes keep pace and as long as the entire marsh can move to higher shore levels or further inland. Provided that it is not constrained by infrastructure, protection works, or other barriers, vertical accretion is likely to neutralize sea-level rise as long as sediment supply is sufficient and horizontal erosion is absent or can be compensated. If not, salt marshes will progressively decline and ultimately disappear. Pethick (1992) has also shown that salt marshes under stable sea level undergo cyclical changes to their seaward boundaries; infrequent high-magnitude storm events erode the edges, while intervening lower-magnitude events allow depositional recovery. An increase in the frequency of storm events as a response to sealevel rise would result in the replacement of such cyclical change by progressive erosion. The sensitivity of certain saltmarsh species to waterlogging and soil-chemical changes also could result in a change in species composition or the migration of vegetation zones (Reed, 1995).

Mangroves grow largely in tidal forests and are characterized by adaptations to unconsolidated, periodically inundated saline coastal habitats. They fringe about 25% of shorelines in the tropics and extend into the subtropics as far north as Bermuda and as far south as North Island, New Zealand. Studies on the effects of sea-level rise on tropical mangrove ecosystems have been primarily of historical nature (reviewed by Woodroffe, 1990; UNEP-UNESCO Task Team, 1993; Edwards, 1995). These studies have shown that extensive mangrove ecosystems became reestablished when sea level stabilized around 6,000 years BP. During the prior rise, mangroves probably survived as narrow coastal fringes, shifting landward with the migrating shoreline. Ellison and Stoddart (1991) and Ellison (1993) have indicated that mangroves in areas of low sediment input in both low-island and high-island settings appear to be unable to accrete vertically as fast as the projected rate of sea-level rise. However, recent evidence from the Florida Keys (Snedaker et