Introduction
"Just constructing dykes along all threatened coasts may, of course, not be a good idea"
(CZMS 1990, page 86)

By the "worst case" scenario, global mean sea-level is expected to rise 95 cm by the year 2100, with large local differences due to tides, wind and atmospheric pressure patterns, changes in ocean circulation, vertical movements of continents etc.; the most likely value is in the range from 38 to 55 cm (Warrick et al, 1996). The relative change of sea and land is the main factor: some areas may experience sea level drop in cases where land is rising faster than sea level. In addition, according to a study by Titus and Narayanan, quoted by CZMS (1992), the statistical distribution of sea-level rise exhibits a marked positive skew (i.e. many average values and some very large ones in only a few locations).

Regarding human settlements, Scott (1996) expresses the view that the impacts of sea-level rise and extreme events are likely to be experienced indirectly through effects on other sectors - for instance changes in water supply, agricultural productivity (Brinkman, 1995) and human migration. Considering, in addition, that it is uncertain whether extreme events associated with oceans will change in intensity and frequency (Venugopalan Ittekkot, 1996; Nicholls et al, 1996), it seems likely that the internal dynamics of human demography, coupled with a series of indirect factors, including health factors (WHO, 1996), may eventually play a dominant part [1].

According to Nicholls (1995) quoted by WHO (1996), the majority of the people that would be affected under the worst scenario live in China (72 million) and in Bangladesh (71 million). Between 0.3% (Venezuela) and 100% (Kiribati and the Marshall islands) of the population would be affected. It is worth noting, however, that population per se receives relatively little attention in the literature as compared, for instance, to natural ecosystems or agriculture.

A disaster results from the impact of an extreme physical event on a vulnerable society or human activity (Susman et al, 1983). Disasters can be quantified and predicted only insofar as the factors of the product "extreme event x vulnerable system" are reasonably well known and quantified. In the specific case of sea-level rise (SLR) and population, only some terms of the equation are known: at the macro level, population growth is affected by the least error, but details of future population distribution, as well as the level of urbanization, are more open to debate, especially as to whether the future concentration of population will coincide with the area corresponding to the large positive skew referred to above. The vulnerable system itself is currently difficult to describe at the global level, for two reasons. First, sufficiently detailed digital maps of elevation, crops and population are not available; second, the future dynamics of the response of coasts, coastal human activities and populations is largely open to debate. As to future impacts around years 2050 or 2100, we are not in a position either to describe with any level of accuracy and confidence what the impacted systems will be like because, inter alia, both the coastal landscape and buildings and infrastructure will adapt gradually in response to the changing environment and the socio-economic driving forces.

The main weaknesses are thus in the sea-level rise predictions, as well as the interactions with human activities.

The general demographic, physiographic and socio-economic setting
Contrary to a common assertion [2] according to which "it is estimated that 50-70% of the global human population lives in the coastal zone" (IPCC 1996b, p. 294), the population is rather land-bound, as illustrated in Table 1 below. The densities given are approximate in that they are based on an assumed total length of the coastline of 100,000 km and on "large, round" continents. Global population density is about 39 persons/km2. In spite of the gross approximations involved in the last column of Table 1, it is clear that population densities are far higher along the coasts than inland. Small (personal communication) indicates the percentages to be 37% within 100 km, and 66% within 400 km.

There are, of course, large local differences. For instance, Sestini (1992; quoted by Zwick 1997), writes that: "the importance of the Mediterranean seafront in relation to the rest of the country varies; as an example, it is relatively less so in Spain, France and Turkey than in Italy, Greece, Albania, Algeria, Israel. In Greece as much as 90% of the population lives within 50 km of the coast and all major industrial centres are coast-related as well as much of agriculture. In Egypt, the Nile delta north of Cairo represents 2.3% of the area of the country, but contains 46% of its total cultivated surface and 50% of its population; the [altitude] belt 0-3 m harbours about 20 % of the population (with Alexandria 3.5 mill., and Port Said 450.000 inhabitants), 40% of industry, 80% of port facilities, 60% of fish production." As a whole, world population, now at 5,880 million, is expected to begin levelling off around 2050 at about 9,300 to 9400 million people (middle estimates of UN 1996a and 1996b), although in some continents, notably Africa, population will probably continue growing at a more sustained pace well into the 21st century. The urbanized population should exceed 60% of the total in 2030, from current values of around 50%.

Figure 1. Recent and future population and urbanization trends (based on data from UN 1996a, 1996b and 1997)

It is also well known that most of the current largest urban concentrations [3] are on the seacoasts (Engelman 1997). The population in the world's 15 biggest cities is projected to be 223 million in the year 2000. It appears that overall urban population trends are not so obvious. While most coastal megacities do grow in size, their share of the total population often remains stable (1%, from 1950 to 2015 in Calcutta and Shanghai), and sometimes decreases (from 7% to 6% in New York in the same period. Also, the percentage of the urban population living in the megacities often decreases (New York: 12% to 7% between 1950 and 2015; Cairo: 35% to 32%; Rio de Janeiro: 14% to 6%; Calcutta: 7% to 4%; Beijing: 6% to 2%; Jakarta: 15% to 10%, etc.), which points to the growth of other urban areas.

Also noteworthy is the fact that many cities suffer land subsidence due to groundwater withdrawal (Nicholls and Leatherman, 1995). This, of course, may be compounded by sea-level rise, the more so since current rates of subsidence may exceed the rate of sea-level rise between now and 2100.

Table 2 below indicates how some socio-economic and physiographic indicators vary among land-locked countries, those with coastlines and the smallest islands, which are members of the Alliance of Small Island States (AOSIS). It is striking that the average Gross Domestic Product (GDP) per capita in landlocked countries is just above that of AOSIS members, and well below the global average. It is also worth noting that the population densities of the small island states are currently intermediate between landlocked and "other" countries, while they should reach a value comparable to "other" countries in 2050. It is unlikely that this will be accompanied by a marked increase in GDP per capita.

In order to examine with more detail the relation between some of the indicators in Table 2 and the "insularity" of the respective countries, we define an "Insularity Index" as the ratio between the length of the coastline (km) and the total land area (km2) that it encloses:

The definition of The "Insularity Index" is, of course, fraught with problems [4] linked with the actual shape of countries, the fractal nature of coasts, and the scale at which it is determined, and the distribution and extent of low-lying areas within each country. It is admittedly a crude index, but meaningful if a consistent method is used to estimate the length of the coastlines. Some interesting links with other variables can be found at the global level. Table 6 lists some typical values of The Insularity Index.

It is obvious, to start with, that Europe and Africa - which represent 50% of all countries and territories - are far less "insular" than the other continents. It is also apparent that the insular character has a strong positive skew and covers 5 to 6 orders of magnitude, to the extent that it can be represented only on a logarithmic scale. The positive skew is clearly visible in Table 3, in which the large difference between median and average is due to the occurrence of a limited number of very high values.

The two figures below clearly show the association between the insular character of countries and population density: less insular countries are generally relatively less populated, which is linked to the fact that landlocked countries are mostly at higher elevations and latitudes (Eurasia) or in semi-arid areas (Africa) where productivity and population supporting capacities tend to be low.

Figure 2. Variation of current population density (1995) as a function of the Insularity Index
Based on 192 countries and territories for which the data are available in UN 1996a and UN 1997 (population data) and in Factbook, 1997, for the computation of the Insularity Index.

In a similar way, and as a consequence of the situation described previously, GDP tends to decrease with insularity, mostly so above a "threshold" of 0.01.

Figure 3a. Variation of 1996 Gross Domestic Product (GDP) and GDP/capita as a function of Insularity Index

Figure 3b. Variation of 1996 Gross Domestic Product (GDP) and GDP/capita as a function of Insularity Index

Based on data from Factbook, 1997.

Notes
1. Direct effects can be assumed to more significant in the ecological sphere (see, for instance Bijlsma, L. 1996).

2. The statement is ubiquitous, with many variants, including "By some estimates, nearly two-thirds of the world's population lives within 100 miles of an ocean, inland sea or major freshwater lake" (Engelman, 1997, p 39), or "Six out of ten people live within 60 km of coastal waters" (IUCN, 1991, p. 150) or "At present, six out of 10 people live within 60 km of the coast" (FAO, 1997).

3. They include, in decreasing order of the projected population in the year 2000: Tokyo, Mexico City, Bombay, Sao Paulo, New York, Shanghai, Lagos, Los Angeles, Calcutta, Buenos Aires, Seoul, Beijing, Karachi, Delhi and Dhaka (UN, 1997).

4. For instance, the Factbook (1997) indicates 14,500 km for China (without Taiwan), but Du (1993), quoted by CGER (1996) uses 18,700 km (including Taiwan). India has 7,000 km in both CGER (1996, quoting Asian Development Bank) and Factbook (1997), while CZMS (1990) indicates 3,280 for the length of the coast "as the crow flies" with a step of 50 to 100 km. The same source applies a multiplier of 10 to obtain 32,440 km of "length of low coast", i.e. the actual length of coast that should be protected, taking into consideration its "micro" structure. For Vietnam, the figures are 3,260, 3,444, 512 and 6,095 (respectively CGER, Factbook, CZMS with and without multiplier). For Japan, the figures 29,751, 34,390, 530 and 3,870 are obtained by the same sources. Regarding the total length of all coasts (excluding Antarctica), we found 642,770 based on a 1:25M map, while the data from the Factbook (1997) add up to 715,917. Annex D2 of CZMS (1990) finds 46,185 km using 50 to 100 km "steps" and 339,185 km of "low coast".