| Abstract | Osnovni cilj disertacije bio je istražiti razvoj temeljnice u mješovitim sastojinama hrasta lužnjaka i običnoga graba, utvrditi bitne odnose koji utječu na njezin razvoj te izraditi modele razvoja temeljnice na razini gospodarskih lužnjakovo-grabovih sastojina u Republici Hrvatskoj, na razini trajnih pokusnih ploha osnovanih u sastojinama hrasta lužnjaka i običnoga graba i na razini stabala na trajnim pokusnim plohama. Modeliranje temeljnice na razini gospodarskih sastojina provedeno je na uzorku od 5060 odjela/odsjeka, ukupne površine 75.948,76 hektara. Modeliranje temeljnice na razini trajnih pokusnih ploha provedeno je na uzorku od 28 trajnih pokusnih ploha, ukupne površine 17,92 hektara. Modeliranje temeljnice na razini stabla provedeno je na uzorku od 1.509 stabala hrasta lužnjaka, 3.954 stabla običnoga graba i 267 stabala ostalih vrsta drveća.Za modeliranje razvoja temeljnice kroz vrijeme, na razini sastojine, odabrane su nelinearne funkcije rasta Levaković III, Gompertz, Chapman-Richards i Hossfeld IV.Razvoj temeljnica po hektaru (hrast, grab, ostalo i ukupna) na trajnim pokusnim plohama modeliran je samo Levakovićevom funkcijom rasta.Kako je razvoj temeljnice mješovitih sastojina kao i razvoj strukture tih sastojina u cjelini, vrlo je kompleksan i dinamičan te predstavlja funkciju dimenzija stabala, vitaliteta stabala, međuvrsne i unutarvrsne konkurencije stabala i produktivnosti staništa, modeliranje razvoja temeljnice pomaknuto je sa razine sastojine na razinu individualnog stabla.Modeliranje razvoja temeljnice na razini stabla provedeno je kao modeliranje godišnjeg prirasta temeljnice stabala na trajnim pokusnim plohama, kroz više faza modeliranja određenih grupama nezavisnih varijabli, jednostavnom i višestrukom linearnom regresijom stepwise postupkom selekcije nezavisnih varijabli.Iz dobivenih rezultata istraživanja proizlaze sljedeći zaključci: Rezultati korelacijske analize pokazali su da se temeljnica po hektaru, kao zavisna varijabla, na uzorku gospodarskih sastojina, vrlo teško može opisati pomoću samo jedne nezavisne varijable. Analizom rezultata, svih dobivenih modela na razini sastojina, može se zaključiti kako nema značajnih razlika između modela. Koeficijenti determinacije kod modeliranja ukupne temeljnice po hektaru, četiriju odabranih funkcija rasta, se kreću od 0,4374 do 0,4401, a kod temeljnice hrasta lužnjaka po hektaru od 0,2276 do 0,2310.IV Kao najbolji model razvoja temeljnice na razini sastojine odabran je model izjednačenja funkcijom Levaković III, na osnovu koeficijenta determinacije i na osnovu činjenice da je Levaković svoju funkciju rasta konstruirao na osnovu istraživanja u domaćim sastojinama hrasta lužnjaka. Razvoj temeljnice (ukupna, hrast, grab, ostalo) po hektaru i na razini sastojine i na razini trajnih pokusnih ploha karakterizira velika varijabilnost vrijednosti temeljnice. Modeli razvoja temeljnice na razini sastojina i na razini trajnih pokusnih ploha imaju sličan oblik, ali kod razine trajnih pokusnih ploha postižu značajno veće vrijednosti. To ukazuje na činjenicu kako su trajne pokusne plohe osnovane u reprezentativnim mješovitim sastojinama hrasta lužnjaka i običnoga graba. Modeliranje razvoja temeljnice na razini sastojine i razini trajnih pokusnih ploha funkcijama rasta, zbog vrlo malo informacija koje opisuju dinamiku razvoja promatranog elementa strukture, je previše statično i jednostavno. Modeliranje godišnjeg prirasta stabala, na trajnim pokusnim plohama, razinom 0 provedeno je jednostavnom linearnom regresijom. Za nezavisnu varijablu odabran je srednji prsni promjer stabla. Koeficijenti determinacije modeliranja ovom razinom su 0,4113 za hrast lužnjak, 0,3592 za obični grab i 0,4199 za ostale vrste drveća. Modeliranje godišnjeg prirasta stabala, na trajnim pokusnim plohama, razinom 1 provedeno je višestrukom linearnom regresijom. Grupu nezavisnih varijabli ove razine karakteriziraju standardne taksacijske varijable. Koeficijenti determinacije modeliranja ovom razinom su 0,4939 za hrast lužnjak, 0,4034 za obični grab i 0,4916 za ostale vrste drveća. Modeliranje godišnjeg prirasta stabala, na trajnim pokusnim plohama, razinom 2 provedeno je višestrukom linearnom regresijom. Grupu nezavisnih varijabli ove razine karakteriziraju varijable izvedene iz standardnih taksacijskih varijabli i opisuju odnose konkurencije između stabala. Koeficijenti determinacije modeliranja ovom razinom su 0,516 za hrast lužnjak, 0,4314 za obični grab i 0,5465 za ostale vrste drveća. Modeliranje godišnjeg prirasta stabala, na trajnim pokusnim plohama, razinom 3 provedeno je višestrukom linearnom regresijom. Grupu nezavisnih varijabli ove razine karakteriziraju varijable izmjerene, izračunate i izvedene na osnovu podataka dobivenih izmjerom krošanja stabala. Koeficijenti determinacije modeliranja ovom razinom su 0,6154 za hrast lužnjak, 0,5687 za obični grab i 0,5825 za ostale vrste drveća. Modeliranje godišnjeg prirasta stabala, na trajnim pokusnim plohama, razinom 4 provedeno je višestrukom linearnom regresijom. Grupu nezavisnih varijabli ove razine karakteriziraju varijable dobivene pedološkim analizama tala trajnih pokusnih ploha. Koeficijenti determinacije modeliranja ovom razinom su 0,6251 za hrast lužnjak, 0,5748 za obični grab i 0,5922 za ostale vrste drveća. Konceptom modeliranja razvoja godišnjeg prirasta temeljnice na razini stabla, promatranom kroz sve razine modeliranja, postupno su uključivane grupe nezavisnih varijabli koje predstavljaju dimenzije stabala, konkurenciju između stabala, vitalitet stabala i varijable krošanja te produktivnost staništa. Na taj način simuliran je razvoj godišnjeg prirasta temeljnice svakog pojedinog stabla. |
| Abstract (english) | The aim of this dissertation is to investigate development of tree basal area in the forest community of the Pedunculate Oak and Common Hornbeam (Carpino betuli-Quercetum roboris Anić 1959 ex Rauš 1969), the factors influencing its development and modelling on the level of managed forest stands, permanent sample plots and trees on the permanent sample plots in Republic of Croatia. Modelling of the tree basal area included 5060 forest compartments/ subcompartments on the level of managed forest stands comprising in total 75,948.76 ha. Modelling of tree basal area on the level of permanent sample plots included 28 plots comprising 17.92 ha. Modelling of tree basal area on the tree level included 1.509 trees of pedunculate oak, 3.954 trees of common hornbeam and 267 other tree species.For the purpose of modelling on the level of forest stand nonlinear growth functions were used: Levaković III, Gompertz, Chapman-Richards i Hossfeld IV.Development of tree basal area per ha for Pedunculate Oak, common hornbeam, other species and in total on the permanent sample plots was modelled according to Levaković III function.The growth of tree basal area in mixed forests is very complex and dynamic process, influenced by several factors such as tree dimensions, tree vitality, the competition between the trees of the same or different species and site productivity. Therefore only modelling of development of tree basal area on the tree level was conducted.For the purpose of modelling of annual increment of tree basal area several scenarios (levels) were constructed with groups of independent variables by using stepwise simple and multiple linear regressions.The major results and conclusions based on this dissertation are as follows: Correlation analysis showed that development of the tree basal area as dependent variable can hardly be explained by using only one independent variable on the stand level. There is no significant difference between the models on the forest stand level. Determination coefficients of four chosen growth functions ranged between 0.4374 and 0.4401, and when modelling tree basal area of Pedunculate Oak per ha it ranged between 0.2276 and 0.2310.VI The best model of tree basal area on the stand level was Levaković III growth function. The reason lies probably in the fact that Levaković constructed his growth function based on analysis in local forest stands of Pedunculate Oak. Development of tree basal area (in total, for Pedunculate Oak, Common Hornbeam, and other species) on the level of forest stand and level of permanent sample plots is characterised by great variability. Models of development of tree basal area on the level of forest stand and level of permanent sample plots were the similar shape, but on the level of permanent sample plots the values were significantly higher. This is the result of establishing permanent sample plots in representative forest stands of Pedunculate Oak and Common Hornbeam. Modelling of development of tree basal area on the level of forest stand and level of permanent sample plots by using of growth functions is too simple and static, due to scarcity of information explaining dynamics of this taxation element. Basic scenario (level 0) for modelling of annual increment on the level of permanent sample plots was conducted by using simple linear regression. Tree diameter at breast height was the independent variable. Determination coefficients were 0.4113 for the Pedunculate Oak, 0.3592 for Common Hornbeam and 0.4199 for other tree species. Alternative scenario 1 (level 1) for modelling of annual tree increment on the level of permanent sample plots was constructed by multiple linear regressions. Independent variables were usual structural variables. Determination coefficients were 0.4939 for Pedunculate Oak, 0.4034 for Common Hornbeam and 0.4916 for other tree species. Alternative scenario 2 (level 2) for modelling of annual tree increment on the level of permanent sample plots was constructed by multiple linear regressions. Independent variables were variables based on usual structural variables which describe competition between the trees. Determinant coefficients were 0.516 for Pedunculate Oak, 0.4314 for Common Hornbeam and 0.5465 for other tree species. Alternative scenario 3 (level 3) for modelling of annual tree increment on the level of permanent sample plots was constructed by multiple linear regressions. Independent variables in this case were variables based on tree crown measurement. Determination coefficients were 0.6154 for Pedunculate Oak, 0.5687 for Common Hornbeam and 0.5825 for other tree species. Alternative scenario 4 (level 4) for modelling of annual tree increment on the permanent sample plots were constructed by multiple linear regressions. Independent variables were variables based on soil measurements. Determination coefficients were 0.6251 for Pedunculate Oak, 0.5748 for Common Hornbeam and 0.5922 for other tree species. Modelling of annual tree increment on tree level in different scenarios was based on gradual introduction of groups of independent variables, such as tree dimensions, competition between the trees, tree vitality, tree crown variables and site productivity. |
