Comprehensive and reliable statistics about forest plantations at the country level are scarce and, where these do exist, are not generally in the public domain. Consequently, it is very difficult to estimate the age-class structure of forest plantations in most countries. Notable exceptions to this are countries that have invested heavily in forest plantations and where forest plantations can be easily distinguished from natural forests (e.g. Australia, New Zealand and Chile). The literature for other countries provides information, of varying completeness and specificity on annual planting rates, harvesting intensities, mortality, time series of total plantation areas, regional inventories, and qualitative resource descriptions. Establishing a global description of plantation age-classes is therefore, necessarily a piecemeal exercise involving a comprehensive literature review for individual countries with data coalesced and manipulated to obtain age-class structures that are consistent with national area statistics.
It is important to recognise that the key purpose of the resource information in this paper is to provide an assessment of plantation resources at the global and regional level. The data analysis consequently takes the form of a "components analysis". Individual countries are treated as components of the whole, and national age-class structures have been formulated independently and then aggregated to regional and global level. The important statistical assumptions in components analysis are that component errors are independent and randomly distributed around a mean of zero. Thus, the greater the number of components assessed, the more accurate should be the aggregated result. The percentage error bounds for the global assessment should consequently be smaller than for its component regional assessments. Similarly, the error bounds for each regional assessment should be smaller than for individual country components. An obvious skewing of the true error distribution occurs if large errors are made in estimating the biggest components. Consequently, efforts to ensure accurate assessment of age-class structures in countries' plantation resources increased commensurate with the size of the resource in each country. To this end, time constraints dictated that only countries with plantation resources exceeding 100,000 hectares could be assessed individually. This constraint still necessitated individual analysis in 65 countries constituting 98.5 of the total global plantation resource. Age-class structures in countries with less than 100,000 hectares of plantations (1.5 percent of the global total) are assumed to conform to the average of each regional structure.
Two important reports establish the total forest plantation area statistics (by country), with which the national age-class structures derived here have been made consistent. Pandey (1997) assesses forest plantation areas for 1995 in 90 tropical and subtropical countries. Pandey reports total reported plantation area, net plantation area, industrial plantation area, annual establishment rate and species composition of the resource for each country. Pandey's "net plantation area" was used as the area base for age-class derivation. Provisional data reported to the UN-ECE/FAO Temperate and Boreal Forest Resources Assessment component for OECD countries, for the year 2000 (i.e. an early draft of what is now UN, 2000) are used as the area basis for temperate and boreal forest countries. National area data for temperate and boreal forest plantations were provided for 39 (mainly European48) countries.
A useful point of departure for tropical forest plantation age-class assessment was the FAO Tropical Forest Resources Assessment Project 1980 (FRA1980). This set of reports provides national plantation age-class structures by species for 75 tropical and subtropical countries as at 1980. It also provided forecasts of age-class structures to 1985. The reports were used to benchmark upper bounds on plantation age-classes greater than 15 years. The reporting systems of both the FRA 1980 and the Pandey report differentiate between industrial and non-industrial plantations and this (sometimes artificial) distinction has been maintained.
The most significant data manipulation was carried out in transforming "gross plantation age-classes" developed from the literature review, to "net plantation age-classes" consistent with the area statistics reported in Pandey (1997) and the TBFRA 2000 component (UN 2000). The methodology used approximates that of Pandey:
"Estimation of the net area, that is, the actual area of the stocked plantations excluding failed, harvested or doubly counted plantations, has been done by applying a reduction factor/success rate derived from inventory or survey of plantations."
The methodology in this paper applies Pandey's reduction factor to both industrial and non-industrial plantations. As Pandey notes, this may not be appropriate since it is likely industrial plantations will be better managed, than will non-industrial plantations.
In this analysis, a "harvesting and mortality function" for each country (where necessary) was selected from a family of exponential functions and applied as weighting factors to derived national "gross plantation age-classes" to result in the targeted net area. The harvesting and mortality functions selected were of the form:
Net forest plantation area = Gross forest plantation area in each age class
in each age-class 1-aX
Where: X = an age value for each five-year age class (such that for age class 46-50 X = 0.0015; 41-45 X = 0.0040; 36-40 X = 0.0065 ...1-5 X= 0.0240. For age-class >50, an arbitrary value of 0.0010 was applied)
a = discretionary variable
The precise value selected for a relied to a certain extent on the consultant's judgement. A general value of a was estimated using the length of rotation of the predominant species in a country and an assessment of the extent of mortality in the country. A large value of a corresponds with short rotation lengths and high mortality (applying greatest weight to most recent planting data). This general value of a was then refined to improve the match of the derived data with benchmark reported statistics.
Figure 29 Example of representative versus true age-class structures
The country age-class data presented in Appendix 3 are, therefore, both derived from the original literature and have been subjected to the adjustment described above. It should clearly be regarded as "broadly representative" of national age-class structures rather than as exact for each age class. The example displayed in Figure 29 demonstrates this point. In Figure 29, the true age-class distribution of plantations for a country is represented by the bar graph. A "representative" age class distribution, which embodies some of the characteristics and magnitudes of the true age-class distribution, is represented by the line-graph. It is this general shape of the data that the estimation process described above seeks to replicate.
48 Non-European countries included in the assessment were: Australia; Canada; Japan; New Zealand; and United States of America.