## Info

Subtropical and tropical woodland

estimate redundancy and the cost of information contained in the text at different levels of description.

Now let us consider a description of the global vegetation pattern (GVP). At the first level of description we have a list of biomes, or vegetation types. In accordance with Walter (1964, 1968) and Bazilevich (1973, 1993)—see also Svirezhev (2002)—the number of different biomes is equal to 30, and they are listed in Table 4.1. The corresponding biome maps can be found in the above-mentioned books.

If we know nothing about these biomes except for their denomination and number, then the natural assumption at the zero level of description (and perception) is that they are absolutely equivalent in the list. If we consider the list as some alphabet with 30 letters, then the information per letter is equal to I(0) = log2 30 = 4.9 bits. However, this is not the case with biomes, since they are not entirely equivalent. Certain biomes occupy large areas, while the areas of others are negligibly small (for instance, biomes no. 9 and no. 28 in Table 4.1), and the production of some biomes is also much higher than the production of others, etc. In other words, in reality, the GVP is a hierarchical structure with respect to different characteristics, many of which are known (see Table 4.2). Therefore, at the next (first) level of description all these properties might be taken into account. For instance, using the data from Table 4.2 we can calculate the relative areas of biomes, pi = 3=i where si is the total area of ith biome.

The information per letter at this level of description is equal to I1-1' = 301 Pi log2 Pi = 4.41 bits. The redundancy of information at this level is equal to R1 = 1 - (4.41/4.9) = 0.1, and its cost C1 = 1/(1 - 0.1) = 1.11. These values are calculated by using the formulas of Section 4.3.

Up to this point we have not taken into account the spatial pattern of global vegetation, i.e. the spatial correlation between different pairs of biomes, which is a very important characteristic of the GVP. For instance, the correlation between tundra and north taiga biomes is very high, but in contrast the correlation between tundra and evergreen tropical

### Table 4.2

Annual net primary production, P (kg C/(m2 year), density of living biomass, B (kg C/m2) and density of dead organic matter, D (kg C/m2, in 1 m soil) for different types of vegetation (biomes); a, biome type; b, biome area ( X 106 km2)

### Table 4.2

Annual net primary production, P (kg C/(m2 year), density of living biomass, B (kg C/m2) and density of dead organic matter, D (kg C/m2, in 1 m soil) for different types of vegetation (biomes); a, biome type; b, biome area ( X 106 km2)

a

b