Stand structure and development stages

The concept of forest development stages has a long tradition in the description of stand structure and structural dynamics in Central Europe. However, the existing approaches for classifying different stages of stand development seem incapable of displaying structural changes in a comprehensive way (Christensen, 2007; Kral et al., 2018). Hence, their explanatory value in the description and analysis of forest stand structure and dynamics seems questionable. Likely, the problem is a strong emphasis on ‘structural master factors’

that determine the development stage and implicitly assume structural uniformity and thus, a collective development at the chosen spatial scale. The proposed development stage index I_DS (chapter 5) is a new approach to objectively quantify the spatial extension of three forest development stages (Initial, Optimum and Terminal) at different scales. It mainly differs from the existing approaches in two aspects: i) It incorporates multiple levels of mixing of life history stages on the observational scale used, and ii) dead trees are not per se attributed to a Decay or Terminal stage, but account for different stages depending on their dimension. Thereby, the I_DS provides information on variable mixing of tree populations in different demographic positions (stages) by only relying on a very confined set of input data. A great benefit is that the mixing of stages can be continuously displayed by ternary graphical plots.

The three development stages had a relatively balanced mean spatial extension (27–39%) in the three virgin beech forests studied, which suggests that they are not far from so called ‘structural equilibrium condition’ at the stand scale. It turned out that the development stages rarely occur in strong areal dominance and there is a high variation in the mixing of stages on circular plots of 500 m² or even smaller (156 m²). This reveals that by ignoring subdominant structures, the existing approaches largely blanked out a very characteristic feature that is highly relevant to future structural development.

Synthesis 6.2

If intense and diverse mixing of stages is rather the rule than the exception, one may argue if the concept of development stages makes sense in these forests. But when the structures found in the Kyjov forest are regarded against the background of gap dynamics, which means going back to the roots of the forest cycle concept (Watt, 1947), there is a lot of conformity.

Canopy gaps occurred, ranging in size from 24 to 4415 m², which shows structural development can be affected at very different scales. Frequent endogenous small-scale disturbances were found to drive the structuring processes on a considerable proportion of the forest area in Kyjov. Tree development supposedly takes very individual pathways here, depending on the time interval until the next canopy opening and its size. Dendroecological studies on beech trees grown in virgin forests strongly support this assumption (Trotsiuk et al., 2012; Nagel et al., 2014; Hobi et al., 2015). As a result, trees in multiple life history stages may occur in direct proximity to one another. In accordance, the degree of small-scale mixing of trees in the Initial (7–39 cm DBH), Optimum (40–69 cm dbh) and Terminal (≥ 70 cm DBH) stages was indeed high in the Kyjov forest with frequent stage overlap within a plot of 500 m² (chapter 5).

In larger gaps, newly establishing cohorts might obtain a higher degree of areal dominance, which is presumably correlated with gap size. However, when exceeding a certain scale, some (rather low) level of structural heterogeneity seems obligatory. The frequency of canopy gaps decreased exponentially with increasing gap size but rare large gaps affected comparably large areas on a spatio-temporal scale. Thus, more extensive patches of less heterogeneous structure could also be expected on a considerable portion of forest area in the Kyjov forest. That a single stage dominated over the others was found on many plots, but their number decreased exponentially with increasing factor of dominance (i.e., the difference between the highest I_DS value and the closest of the respective other two I_DS values). There were no plots with an exclusive presence of a single stage.

Altogether, the structures encountered on the inventory plots in Kyjov, as described by means of the I_DS, are very consistent with the trends described for structural development driven by gap dynamics in this forest. The development stage index (I_DS) thus seems an appropriate tool to comprehensively describe and analyze forest structure in terms of forest dynamics in this forest.The general functioning of gap induced structural processes, i.e., small gaps promote vertical heterogeneity while larger gaps may cause more homogeneously structured patches, can be anticipated to be roughly similar in other beech-dominated virgin forests. Thus, it seems valid to interpret forest dynamics by analyzing stand structure applying the I_DS in these forests as well.Looking at the ternary graphical plots (chapter 5, Fig. 5.3) this notion would suggest that, despite a relatively balanced mean areal presence of the three development stages in the

Stand structure and development stages 6.2

studied forest stands, there are also some differences in forest dynamics, as the stands revealed different patterns in plot locations. Assuming that endogenous disturbances cause small-scale heterogeneity and with increasing disturbance intensity (i.e., increasing gap size) the resulting structure becomes increasingly homogenized, a research plot’s distance to the center of the ternary plot (that represents maximum heterogeneity) could be interpreted as a gradual indicator for the intensity of past disturbance. In the long-term absence of more intense exogenous disturbances, endogenous small-scale disturbances would drive structural development that should tend towards positions close to the center (attractor). Consequently, the pattern is largely determined by the frequency and intensity of exogenous disturbances.

According to this interpretation scheme, Havešová and Stužica would have experienced more severe disturbances than Kyjov, as in these stands there are more plots at rather eccentric locations. The clumped occurrence of sample plots, e.g. left of the center in the ternary plot for the Havešová forest (Fig. 5.3), might indicate that a disturbance event caused damage at the stand scale.

So far, the interpretation of the ternary plots remains speculative. Similar structures may originate from different pathways of stand development. However, at least for Kyjov, the interpretation seems substantiated by the results on gap dynamics in this forest (chapter 2 and 3). There are two major issues regarding the validation of the development stage index as a conclusive tool for the interpretation of forest dynamics: i) The application of the I_DS on repeated inventory data would inform about structural developmental pathways and could provide information on structural self-organization in beech-dominated forests. Further, ii) the I_DS could be adapted to be applicable on continuous stand structural datasets (moving window, focal filtering). This would inform on spatial pattern in stand structure and allow for mapping of the horizontal texture of development stages. A combination, i.e. repeated inventory data from a continuous area (several hectare), would be most suitable.

Yet the application of the I_DS is limited to our test systems, as the index relies on reference values for stand density that were specifically generated for these forest stands. However, the impact of site conditions, and also that of different tree species, on maximum stand density is consistent across tree-dimensions and causes parallel shifts towards higher or lower densities in even-aged stands (Reineke, 1933; Pretzsch and Biber, 2005). This indicates that the application of the reference values to other forests would only affect the absolute values of the stages I_DS, but not the relative proportions in a plot or in a stand. The general applicability of the reference values, and a potential adjustment of these, need to be evaluated to allow for a broad implementation of the I as a tool in the analysis of forest structure and dynamics.

Synthesis 6.3

6.3 Implications for forest ecosystem conservation, restoration and