This publication is largely based on growth-site studies carried out on important tree species in Dutch forests during the period 1950-1991. Two of the ways the data yielded by these studies can be used were investigated. The first aimed to identify important correlations between growth and site properties. The second used combinations of site properties in an attempt to predict site index; this is important for the proper choice of tree species, or if previously unforested land is to be afforested (but only if the site properties do not change intrinsically in response to changes in environmental conditions).The investigations centred on the Dutch forest as it has evolved since the mid- nineteenth century. The pioneer tree species was Scots pine, later partly replaced by other species. Chapter 1 presents a short description of the history of heathland afforestation and examines the background of growth-site research. In Chapter 2 a description is given of the character of growth-site research and the methods commonly applied. A critical discussion of methods and interpretation is included.Growth-site research in Northern America and in northwestern Europe is discussed in Chapter 3, in relation to the methods applied and relevance for the interpretation of relationships established for forest stands in the Netherlands. Attention is paid to the international literature on the study of growth-site relationships. Forestry in the Netherlands is distinctive, because it is a rather young type of land use, which occupies a small area and was until recently mainly restricted to poor soils. The range of meteorological variables in the Netherlands is small, and therefore the interpretation of results of growth-site studies must rely partly on studies in neighbouring countries. This chapter also describes the history of growth-site research from the nineteenth century to the present.Chapter 4 gives a comprehensive description of the field and laboratory methodology applied in Dutch growth-site research from 1950-1991. This methodology is compared and checked with procedures published in Dutch and international literature.The essence of this publication is contained in chapters 5, 6 and 7. These chapters describe the results of growth-site studies on Douglas-fir (an "exotic" tree species), European ash (a "native" species), and 'Robusta' poplar (hybrid of the "exotic" tree species Populusdeltoides and the "native" tree species Populus nigra ). 'Robusta' poplar was chosen as being representative of poplar clones grown in the Netherlands.In the research, site index values (derived from top height - age relations for the three tree species) were grouped according to various factors (forest area, soil parent material, groundwater class, soil unit, former land use, and vegetation type), and tested for significant differences, by analysis of variance. Relationships between site index and single predictor variables (representing soil chemical properties, water supply and meteorological conditions) were investigated by application of linear, quadratic, logarithmic and exponential regressions, and checked for plausibility of the relation ("identification" of variables, as described in section 2.1). The main soil chemical variables tested were pH-KCI, P-total, N-total, N org (N expressed as concentation of organic matter), and incidentally available K (K-HCl 0.1M) and Mg (Mg-NaCI 0.5M). Water supply index was estimated as available soil moisture in the rooted zone (pF 2.0/2.2 - pF 4.2) for soil profiles with a deep water table, and as available soil moisture (pF 1.7/2.0 - pF 4.2), to which is added water supplied by capillary rise, for soils with a shallow water table. Meteorological variables were derived from the Climate Atlas of the Netherlands (1972), and included mean temperature, precipitation and atmospheric moisture deficit (precipitation - 0.8 * Eo) for the months May up to and including September, mean January temperature, mean annual wind velocity, mean annual number of frost days, and mean number of annual sunshine hours (for 'Robusta' poplar only).Variables and factors which individually explained significantly the variance of the site index, were combined by multiple regression analyses to produce a simple prediction model for the site index, assuming that there were at least direct relationships ("causality") between variables/factors and site index. These models were compared with the results of other growth-site investigations, with literature data, and with regression models derived from soil survey data.Data from two periods (1959/60 and 1984-1991) were examined to establish growth site relationships for Douglas-fir. This enabled the impact of changing environmental conditions in the Netherlands during the sixties, seventies and early eighties to be assessed. The results demonstrated an important shift in the relationship between soil P supply and site index of Douglas-fir. The 1959/60 data revealed that P-total contributed significantly to the variance of the site index, but this effect was absent in the 1984-1991 data.Chemical and physical soil variables are often used to characterise forest soil properties, and to predict forest growth. There is a different method, which uses forest vegetation to indicate forest soil properties. Bannink, Leys & Zonneveld (1973) developed a variant of this method for Dutch conifer forests on sandy soils, calibrating it by applying the site index of Scots pine stands. Bannink and his collegues contended that their method reflected soil fertility levels, but they did not check this thoroughly. Chapter 8 describes the relationship between classes of this vegetation system and some forest site properties (water supply, soil chemical characteristics). The data were derived from growth-site studies performed during the period 1959-1990 in stands of Douglas-fir, Norway spruce, Japanese larch, Scots pine, beech, pedunculate and sessile oak, and 'Robusta' poplar. The vegetation of these stands was classified acoording to the system as developed by Bannink et al.Site properties for vegetation types in "dark" and in "light" forests were compared and tested for significance. A multivariate method (discriminant analysis) was used to relate vegetation classes to a combination of site properties. Vegetation types in "dark" forests (beech, Douglas-fir, Norway spruce) reflected differences in P-total and water supply, and between tree species themselves, but not in pH-KCI. In "light" forests, vegetation type differences were related to P-total, N org , water supply, and pH-KCI, but the relationship with pH-KCI was only significant if sites included both calcareous and non-calcareous soils. In stands of Scots pine and pedunculate oak on acid soils, vegetation types did not show significant differences in pH-KCI values. From this it is concluded that the system of vegetation types as developed by Bannink, Leys & Zonneveld (1973), currently applied in routine forest soil surveys, does indeed reflect differences in soil fertility (indicated by P-total and N org ) and in water supply to some extent, but not in pH-KCI, as far as acid soils (pH-KCl < 6.5) are concerned. The tree species itself also has some impact on the vegetation type on otherwise identical soils.Relationships between site properties and the growth of various tree species as derived from the results of growth-site research can be applied for qualitatively or quantitatively assessing the suitability of land for a specific land use type, in this context, various types of forestry. Chapter 9 presents an overview of the history of soil suitability assessment and land evaluation for forestry in the Netherlands, a description of the ALES automated land evaluation system (Rossiter & Van Wambeke 1993), and some of its applications for various forestry land use types (land evaluation for Scots pine based on soil requirements, suitability for poplar cultivation in short and in long rotation). The reason that interest has grown in land evaluation for forestry in the Netherlands is that EU agricultural policy is changing and, in response, land use is switching from agricultural production to, for example, forestry (in the Netherlands: biomass plantations in short- rotation, medium-rotation of poplar and Norway spruce) and nature conservancy. In the Netherlands, land evaluation for poplar in short rotation is based on relationships between dry matter production of fast growing poplar clones and site properties (soil and climate variables), established in field trials and growth-site investigations.On the basis of knowledge about growth-site relationships, and on the basic tenets of ALES, Chapter 9 assesses and discusses the practicability of ALES for Dutch forestry, given that the aim of the land evaluation may be national or stand level. If suitability for forestry must be assessed in qualitative terms, a system like ALES, which renders suitability in qualitative classes, must be capable of being used to assess land capability in a broad sense, i.e. to reject land that is unsuitable for a specific type of land use. ALES can also be used in a quantitative way if site class is a land quality, and the relationship between site class and land characteristics is known. However, it appears that ALES does not make optimal use of knowledge of growth-site relationships in forestry. The growth and yield of forest on land considered suitable for these land use types, can better be predicted by specifically quantitative methods.