La lecture en ligne est gratuite
       Andrius Kulieis    HOMOGENEITY AND EDGES DEVELOPMENT OF FOREST COMPARTMENTS      Summary of doctoral dissertation Biomedical sciences, Forestry (14 B)         
K AUNAS, 2005
The doctoral dissertation has been accomplished during the period of 2000-2004 at the Lithuanian University of Agriculture Scientific adviser: Prof. dr. habil. Romualdas Deltuvas (Lithuanian University of Agriculture, Biomedical sciences, Forestry  14B);   The doctoral dissertation will be defended at the Lithuanian University of Agriculture at the Board of Scientific Trend in Forestry: Chairman:  Prof. dr. habil. Juozas Ruseckas (Lithuanian Forest Research Institute, Biomedical Sciences, Forestry 14 B); Members:  Dr. Algirdas Augustaitis ( Lithuanian University of Agriculture, Biomedical sciences, Forestry  14B); Prof. dr. habil. Romualdas Deltuvas (Lithuanian University of Agriculture, Biomedical sciences, Forestry  14B); Dr. Jurga Motiej ū nait ė (Institute of Botany, Biomedical sciences, biology 01 B);  Prof. dr. habil. Vida Stravinskien ė  (Vytautas Magnus University, Biomedical Sciences, Ecology and Environmental Sciences 03 B); Opponents:  Prof. dr. habil. Romualdas Juknys ( Vytautas Magnus University, Biomedical Sciences, Forestry 14 B);  Assoc. prof. dr. Juozas Maeika (Lithuanian University of Agriculture, Biomedical Sc., Forestry 14 B).     Summary of doctoral dissertation was distributed on May 30, 2005. A copy of the summary of doctoral dissertation is available at the libraries of the Lithuanian University of Agriculture and the Lithuanian Forest Research Institute.
       Andrius Kulieis    MIKO SKLYP Ų HOMOGENIKUMAS IR J Ų RIB Ų  FORMAVIMOSI D Ė SNINGUMAI     Daktaro disertacijos santrauka Biomedicinos mokslai, mikotyra (14 B)               
K AUNAS, 2005
Disertacija parengta 2000  2004 metais Lietuvos em ė s ū kio universitete.
Mokslinis vadovas:  Prof. habil. dr. Romualdas Deltuvas  (Lietuvos em ė s ū kio universitetas, biomedicinos mokslai, mikotyra 14 B);  Mikotyros mokslo krypties tarybos sud ė tis: Tarybos pirmininkas: Prof. habil. dr. Juozas Ruseckas (Lietuvos mik ų institutas, biomedicinos mokslai, mikotyra 14 B);  Nariai:  Dr. Algirdas Augustaitis (Lietuvos em ė s ū kio universitetas, biomedicinos mokslai, mikotyra 14 B);  Prof. habil. dr. Romualdas Deltuvas  (Lietuvos em ė s ū kio universitetas, biomedicinos mokslai, mikotyra 14 B);  Dr. Jurga Motiej ū nait ė (Botanikos institutas, biomedicinos mokslai, biologija 01B);  Prof. habil. dr. Vida Stravinskien ė (Vytauto Didiojo universitetas, biomedicinos mokslai, ekologija ir aplinkotyra 03 B);   Oponentai:  Prof. habil. dr. Romualdas Juknys (Vytauto Didiojo universitetas, biomedicinos mokslai, mikotyra 14 B);  Doc. dr. Juozas Maeika (Lietuvos em ė s ū kio universitetas, biomedicinos mokslai, mikotyra 14 B).
Su disertacija galima susipainti Lietuvos em ė s ū kio universiteto ir Lietuvos mik ų instituto bibliotekose.  
BRIEF CHARACTERISTICS OF THE WORK ------------------------------------------------------------------------------------6  
1. LITERATURE ANALYSIS --------------------------------------------------------------------------------------------------------7  
2. STUDY OBJECT. METHODS -----------------------------------------------------------------------------------------------------9  2.1. Study aim, objectives -----------------------------------------------------------------------------------------------------------9  2.2. Study object ----------------------------------------------------------------------------------------------------------------------9  2.3 Methodical peculiarities ------------------------------------------------------------------------------------------------------- 10  2.3.1 Efficiency estimation methodics of object division into compartments ------------------------------------------- 10  2.3.2. Study methods on forest compartment homogeneity ---------------------------------------------------------------- 10  2.3.3. Study methods on the changes of compartment boundaries -------------------------------------------------------- 11  3. STUDY RESULTS------------------------- ---------------- 12  ---------------------------------------------------------------- --------3.1 Factors preconditioning the efficiency of forest compartment delineation --------------------------------------------- 12  3.2 Homogeneity of forest compartments --------------------------------------------------------------------------------------- 13  3. 3. Stability of forest compartment boundaries according to inventory data of different years------------------------- 17  4. CONCLUSIONS ------------------------------------------------------------------------------------------------------------------- 20  
PUBLICATIONS ON THE THEME OF DOCTORAL THESIS --------------------------------------------------------------- 21  
 INFORMATION ON PHD STUDENT -------------------------------------------------------------------------------------------- 21
REZIUM Ė ------------------------------------------------------------------------------------------------------------------------------ 22  IVADOS --------------------------------------------------------------------------------------------------------------------------- 22  TRUMPOS INIOS APIE DOKTORANT Ą ----------------------------------------------------------------------------------- 23  
BRIEF CHARACTERISTICS OF THE WORK   Relevance of the problem. Forest compartment under the conditions of intensive forestry is the main unit of management and inventory (Antanaitis, 1961; Franklin, 2001). Regular, every 10 years, standwise forest inventory requires great expenditures on compartment delineation and their depicting in cartographic material. Different executives, guided by the same standards, but without objective forest compartment delineation methodics, may divide one and the same object into compartments in different ways (Rutkauskas, Kenstavi č ius, 1978). Therefore, the variability of forest compartment boundaries is artificially increased. Under the absence of clear boundaries between compartments in the forest and considering that the whole forest is homogeneous, compartment characteristics are usually ascertained ignoring their peculiarities at the edges. This comprises preconditions for the appearance of systemic forest inventory deviations ( Антанайтис  и  др ., 1975; Kulieis, Kasperavi č ius, 2000; Brassel, Lichke, 2001). The accuracy of cartographic material and the influence of deviations on estimation reliability of forest resources are not evaluated (Gemmell et al., 1991; Fish, 2000; Franklin, 2001). Compartment edges, occupying in the forests of the country almost one third of the whole area of stands, are important not only from inventory, but also from forest stand growing viewpoint. What are wood increment losses, or what additional increment may be obtained due to differing growth of stands on the edge of forest compartments. Productivity of trees growing at the edge is a very important factor, optimizing the size and perimeter of forest compartments. Compartment edge, the characteristics of growing in them trees and stand formation peculiarities are not sufficiently studied under the conditions of intensive forestry not only in Lithuania, but also in general (Oliver & Larson, 1996; Saladis, 1998; Deltuva, 1999). Aim and objectives of the work. The work was aimed to study the structure of forest compartments and the interdependence of changes in their edges. Seeking to achieve the aim, the following objectives had to be solved: - Estimate delineation efficiency of compartments formed during forest inventory. - Estimate homogeneity of forest compartments. - Estimate the stability of forest compartment boundaries. Scientific novelty. 1. Within stand structure edge, pre-edge and interior zones were delineated, which differ in tree growth conditions and stand yield. 2. Factors preconditioning the stability of compartment boundaries were estimated. 3. The efficiency of forest compartment boundary delineation in standwise forest inventory was estimated. Scientific and practical significance. Elaboration of the methodics of   forest division into compartments is a  significant input  into the improvement of forest compartment delineation methodology.  The worked out methodics allows to estimate objectively the efficiency of forest division into compartments, applying different surveying technologies, to organize the training of taxators on the basis, to specify forest inventory results.  The revealed regularities of forest division into compartments and compartment delineation preconditioning factors lead to efficient forest inventory and the improvement of mapping work. Analysing the structure of forest compartments, zones characterized by different stand parameters and their formation progress were singled out. Sparsely spaced trees, growing at the edge, had a more variable growth space, greater increment accumulation possibilities as compared to tree growth, yield formation and accumulation possibilities in the interior part of compartments. The ascertained regularity essentially supplements already known regularities of stand yield formation, depending on initial growth conditions and tree growth in dense groups or nearby forest glades and gaps. The result of this study will help to elaborate improved sampling design, attain higher accuracy of inventory, estimate edge influence on stand yield, stability, biological diversity, take all the factors into account planning forest compartments. The ascertained variability of compartment boundaries due to random shifts of the main linear objects in cartographic material identifying compartment boundaries during repeated inventories allowed to reveal the main directions of the work seeking greater stability of boundaries, precision of cartographic material: to create a permanent geodetic network, to adjust as much as possible the boundaries of forest compartments with the boundaries of habitats, other naturally existing linear objects, avoid artificial division of compartments by silvicultural measures, organize continuous training of taxators. The analysed variability of compartment boundaries and preconditioning it factors, the ascertained accuracy of cartographic materials in repeated inventories, peculiarities and regularities of stand formation on compartment edge provide significant preconditions improving forest inventory, elaboration of cartographic material and stand formation.
1. LITERATURE ANALYSIS  The idea of standwise forestry, born in the nineteenth century (Antanaitis, 1961), at present has become global, universally accepted (Franklin, 2001). Carrying out forest inventory, it is important to estimate, how long and under what conditions the boundaries of a forest compartment remain stable, or may change (Michalak, 1996). This preconditions the necessity during each forest inventory, every 10-15 years, to control the boundaries of compartments, to specify them or to record as unchangeable. Forest compartment only in exceptional cases may incorporate more than one stand. Several very small stands, where management is similar and there is a possibility to apply the same inventory methods both from economic and inventory viewpoint, may be joined into one compartment. Due to this reason, the concept of a forest compartment is rather often, as well as in this work, analysed along with the concept of a stand. Compartments with a commonly the same vegetation cover in geobotany are called phytocoenoses, in the forest  forest phytocoenoses (Kairi ū ktis et al., 1979). Forest phytocoenose is a community of 5 f  orest plants, formed under certain environmental, natural selection and competition for survival conditions ( М e лехов , 1980). Boundary or edge is a frequently occurring phenomenon in forest landscape, formed naturally or in the result of management (Fish, 2000). Boundary may exist between forest and non-forest area, as a formation of forest and field. A boundary may as well be found in the middle of a forest due to different development classes (mature  young stand), between different species, different management objectives (commercial and strict reserve) (Pauliukevi č ius, Kenstavi č ius, 1995; Deltuva, 1999). Stand boundary is the division of a forest, separating different groups of trees by their density, age, species, growth rate, spatial distribution (Oliver and Larson, 1996; Marozas, Deltuva, 2002). In Lithuanian conditions growth peculiarities of trees growing in edges with openings (Deltuva, 1999), growing nearby forest glades and gaps (Saladis, 1998) were studied. No studies have been carried out on the growth and formation peculiarities of stands growing at the edge of adjacent compartments.  Boundary identification is an important task determining the territory occupied by a compartment, describing each compartment, as well as carrying out instrumental inventory, during which a boundary has to be marked in locality. In forest inventory problems are caused by plots, occurring on compartment boundaries ( Антанайтис  и  др .,  1975;  Kulieis, Kasperavi č ius, 2000; Brassel, Lichke, 2001). Delineation of forest compartments, accuracy of their depicting in maps and stand inventory are closely interrelated (Franklin, 2001). Due to boundary influence, two types of errors are differentiated. These are errors related to inaccurate area estimation and insufficient stand edge representation generally evaluating the compartment (Fish, 2000). Inventory plots, ocurring on the boundary or close to it, practically are discarded or transfered further from the boundary. In such cases errors are inevitable both estimating areas and stand characteristics (Kulieis, 1996).  The stability of forest compartment boundaries or their changeability depends on stand development peculiarities. Stand development is a part of stand dynamics, related to the changes in stand structure over a certain time (Oliver ir Larson, 1996). Boundary, as it is common in forest mapping, is shown as a line. However, quite often an edge is interpreted as a belt, incorporating both stands or a part of both of them, comprising larger-dimensioned trees, greater number of tree species, tree species able to grow under conditions of increased shading (Wales, 1972; Ranney et al., 1981; Neumann, Starlinger, 2001; Tomppo, 2002). The accuracy of forest maps is usually analysed distinguishing in them the most important elements (Weir, 1996; Mowrer et al., 1996). These are exterior boundaries, separating forest areas from non-forest ones, permanent objects, i.e. block lines, roads, rivers and other cartographically important lines and changeable stand boundaries. Dividing an object into forest compartments, random variation of boundaries is faced with. Different taxators, analysing aerial photos of the same object, applying the most advanced forest compartment delineation technologies (Kulieis, Mozgeris, 2000) almost always single out com 6 partments of varying configuration and different number (Rutkauskas, Kenstavi č ius, 1978; Kulieis et al., 2003). Attention to accuracy estimation of forest maps in many countries is paid insufficiently (Gemmell et al., 1991). Recently conducted studies allow to conclude, that precision of stand boundaries depicting in European forest maps comprises 210m (Weir, 1996), in Canadian forest maps  1025 m (Leckie, Gillis, 1995). Accuracy estimation of digital forest maps during the whole history of forest management has been changing from the simplest looks well to the precise identification of general points in locality and forest map, as well as elaboration of error matrix (Congalton, 2004). At present many methods are suggested for the estimation of accuracy of cartographic material (Mowrer, et al., 1996; Quintanilha, Rodrigues, 1996; Owens, McConville 1996). The most important requirement for a reliable method is an accurate positioning of objects, which are easily recognized on the map and in locality (Lodin, Skea, 1996). Literature analysis shows, that the validity and objectiveness of compartment delineation methods is insufficiently analysed in scientific works. Forest compartments are frequently singled out visually, subjectively ascertaining the boundaries. It is important to elaborate methodics of objective forest territory division into forest compartments, guided by the aim to have possibly more homogeneous inside and, as much as possible, differing from each other compartments. Insufficiently are analysed the questions of stability of compartment boundaries and their dynamics, as well as the need during each forest inventory to check and under necessity to change the boundaries of compartments. The accuracy of forest cartographic material is not analysed. The properties of stands growing on the edge of
compartments are not ascertained, as well as their differences compared to stands in the interior part of a compartment. It has not been answered, whether a stand formed at the edge of a compartment is a mixture of two stands, or it is an inseparable part of a stand with peculiar to the edge traits. What are the traits? To answer the questions, growth and formation peculiarities of stands growing on forest edge were analysed, as well as factors preconditioning these peculiarities, the dynamics of forest compartment boundaries and the accuracy of forest cartographic material have been evaluated.  For the studies, data of research and inventory plots were used. To estimate stand structure and process the data, regression, dispersion and other methods accepted in mathematical statistics were applied. To automatize the processing of cartographic material, GIS data bases formation principles and corresponding software (Arcview, Arcinfo) as well as methods were used.  
2. STUDY OBJECT. METHODS  2.1. Study aim, objectives The aim of the work was to study the structure of forest compartments and the interdependence of changes in their edges. To achieve the aim, the following objectives had to be solved: 1. Estimate delineation efficiency of compartments formed during forest inventory. 2. Estimate homogeneity of forest compartments. 3. Estimate the stability of forest compartment boundaries.  2.2. Study object Study object to determine delineation efficiency of forest compartments and to estimate stand inventory accuracy, singled out during forest inventory, was established in Dubrava forest. In Dubrava forest an area of 241 ha consisting of 7 parts was selected, characterized by a different management regime, site conditions, stand species composition, age, forest damages of different level in 1993  1997. In this object were evenly located and measured 1945 angle count plots (Fig.1,2). Study object to ascertain forest compartment zonation and growth peculiarities of pine stands regenerated in clearcut areas was chosen in imoni ų pine wood (Anyk č i ų reg., Fig.1,3). There, in 19201940 prevailed mature pine stands, felled in narrow strips, which is testified by forest inventory and management documents of 1936 saved by the foresters of Sv ė dasai forest district. The study was based also on forest management data of 1936, 1974, 1984 and 1998 as well as delineated and measured 137 square 100 m 2 size research plots. Seeking to estimate stand growth and yield formation peculiarities, in different parts of forest compartments of the main tree species data of permanent inventory plots singled out during national forest inventory were used. The plots represent all the forests of Lithuania (Fig.1). For the study, 4318 permanent inventory plots were measured in 19982002 in pine, spruce, birch and black alder stands among which 1787 are divided into two or more sectors, and at least one represents pine, spruce, birch or black alder stands. The study also used angle count plots, singled out for each sector at 20 m distance from the plot center. Study object to analyse the changeability of stand boundaries and their mapping accuracy was selected in Dubrava forest on an area of 1828 ha. For the study, cartographic material of five forest inventories was used, i.e. forest maps M1:10 000 and inventory descriptions of compartments. The coordinates of intersection points of s: 1 block lines were measured using GPS PRO XRS receiver with ±12 m 4F ig .D1.u Lbroacvaati foonr esscth, e2 m- e oifm sotnuid ų y  poibnjee cwtood,, accuracy. These measurements were used to combine cartographic 3  all country forests material prepared in different forest inventory years elaborating a corresponding digital data base. The whole study material was collected by the author. Permanent plots of NFI he was measuring in 1999.  
2.3 Methodical peculiarities  2.3.1 Efficiency estimation methodics of object division into compartments The essence of objective forest tract division into compartments comprises variance minimization of different inventory indices in the analysed object. Dividing an object into compartments, a solution ensuring a possibly greater difference of adjacent compartments according to mean values of analysed features, maximally decreasing variance of the means inside compartments, is sought. For the purpose the whole study object is divided into equally sized plots, parcels or pixels (Fig. 2). To estimate the efficiency of object division into compartments, the feature, that variance of the estimated object is equal to the sum of variances of all parts of the object, is used. Comparing variance of the object after its division into compartments with its variance before division, the efficiency of object division into compartments is estimated: 1 σ 2 a 2 ft div . E = − er , σ before div .  (1) Fig.2. Location scheme of research k ect           σ a 2 fter div = σ x 2 j × n j , (2) plots in the study obj . j = 1   σ 2 σ 2          before div . = x × n , (3) here: σ 2 xj - variance of x feature in j compartment, n j  number of plots representing compartment, k  n  total number of plots per object, n j = n , j = 1 σ 2 x - variance of x feature in the object Seeking comparability of several variants of forest compartments singled out in the same object, efficiency of each of them is corrected taking into account average compartment area.  2.3.2. Study methods on forest compartment homogeneity   Boundary of the chosen to study forest compartment during its formation is identified in orthophotomap and theoretical coordinates of turning-points of the boundary are calculated. To identify boundaries in locality, GPS receiver is used. To study stand structure, to delineate different zones in the compartment as well as to estimate characteristics of growing in them trees, 10 m wide strip plots are used. They are located perpendicularly to compartment boundaries. The strip plot is divided into 10x10m size plots (Fig.3). Each studied forest compartment has to contain 45 strips, while each strip 57 plots. In each singled out plot trees are inventoried according to the rules of National Forest Inventory (Kulieis, Kasperavi č ius, 2000). Tree measurement data of each plot are processed according to a unified processing methodics of dendrometric information (Kulieis, 1985). The data of strip plots, belonging to one compartment, are analysed applying variance analysis and the method of difference in indices between adjacent plots ( Гмурман , 1972). According to sampling scheme of National Forest Inventory (Kulieis et al., 2000, 2003), permanent inventory plots equally represent all forest compartment zones: interior  circular complete plots, edge  sectors of circular plots and pre-edge  angular count plots allocated at 20 m distance from the plot centre. For each forest compartment zone mean tree diameter, height, stand volume per hectare, stocking level and standard deviations of all these indices are estimated. The dependance of all stand characteristics on age is determined. To Fig.3. Identification of compartment boundaries according to forest inventories of 1936 and 1998 and allocation of research strip plots in the object
compare mean characteristics of separate zones in each 10-year-long age class under probability 0,683, confidential intervals are calculated ( Гмурман , 1972; Кокрен ,1976). 2.3.3. Study methods on the changes of compartment boundaries   Scanned  forest maps of  19581988, using intersection coordinates of block lines and forest inventory GIS data base, were coordinated and vectorized. Compartments in inventory material were identified according to their numbers (Fig.4). To estimate depicting variability of block lines in cartographic material, a 400x400m density network was worked out. For each intersection of the network and block line, an interval, where changes the position of block lines recorded during all inventories as well as deviation of the position of block lines in all inventory years from the average position, was ascertained. Estimating the stability of compartment boundaries, three compartment boundaries variation belts of different width were used - ±2,5, ±5, 0 and ±7,5 metres. ±7,5 m wide belt was accepted by us as the main one. It corresponds to the accepted in forest inventory compartment boundary ascertainment precision ±5m and determined in our work changeability of block line Specifying of the intersection position of block lines by GPS receiver position. Coinciding under probability 0,95 were considered boundaries, when Forest inventory data of 2002 formed around them belts of a certain width at least touch each other, i.e. the Forest inventory data of 1988 Forest inventory data of 1978 boundaries themselves at the point of measurement are away from each other Forest inventory data of 1968 Forest inventor data of 1958 not more than 5, 10 or 15 metres. It was accepted, that compartment boundaries coincide accidentally when they intersect each other Fig.4. Changeability of stand perpendicularly. According to the length of coinciding or at least partially boundaries in the run of five inventories overlapping tc obmupnadrtarmieesn ta nbdo uthned adreiegsr eien  odfi tfferent inventories, the stability of and specifying of the intersection heir continuity from one inventory o GPS receiver tcoo amnpoatrhtemr eisn ascoertained.  psition of block lines by