5.13.1 Exploiting Scattered Long-Term Experiments for Assessing Stand Growth, Resistance, and Climate Smartness by Pooling and Overarching Evaluation of Data
Despite the credo that substantial information and knowledge require big datasets, overarching pooling and evaluation (Liang et al. 2016; Hilmers et al. 2019; Pretzsch et al. 2019) are still an exception in forest ecology or silviculture. For instance, long-term experiments on thinning were established and treated in a way standardized by, e.g. IUFRO (Becking 1953; Johann 1993); however, their evaluation is often restricted to the country or even to the regional level. The tremendous efforts and costs of maintaining long-term experiments may restrain the respective institutions from offering and exchanging their valuable data before having it exploited for their own purposes and publications. However, the evaluations at the state or national level are often of limited value. Samples and evaluations restricted to the state or country level may cover just a small portion of the natural distribution and niche of a tree species. Thus, general and reliable knowledge and relationships between growth and site conditions require trans-geographical evaluations. The broader the range of covered site conditions, the sample size, and the genetic diversity, the more robust and overarching the results for understanding, modelling, and scenario analyses and predictions (Liang et al. 2016; Zeller et al. 2018; Pretzsch et al. 2020a, b).
On the one hand, overarching networks are relevant for monitoring anthropogenic impacts on forest ecosystems and tree species. Local observations may indicate changes, but not their dependency on large-scale environmental conditions, and any shifts in the distribution ranges. On the other hand, overarching networks may contribute to adaptation and mitigation of forest ecosystems to climate change; overarching evaluations and robust growth-site relationships have gained higher importance under climate change. Many studies contribute to better understand the suitability of a given tree species under the expected future climate in, e.g. Central Europe by studying its performance in the north or south, where the species faces already at present the climate predicted for Central Europe in the future (Hanewinkel et al. 2014). This kind of trans-geographical studies is important for detection and remedy of climate changes along latitudinal gradients in the lowlands. And they are even more important along elevation gradients in the mountain forests, where environmental changes may happen in much more narrow space and in shorter time spans with severe consequences for a plenitude of socio-economic forest functions and services (Seidl et al. 2019; Hilmers et al. 2020).
5.13.2 The Information Potential of Long-Term Versus Inventory Plots
The pros and cons of long-term experiments compared with temporary plots have been questioned repeatedly (Gadow 1999; Nagel et al. 2012). Often, long-term experiments have been abandoned in order to cut costs. The common reasons for giving up long-term experiments are that forest areas with long-term experiments have to be left out from regular forest operations and that their establishment and survey are costly. In addition, it hardly fits to the contemporary funding organizations and the zeitgeist that it requires a couple of years for getting the first results.
Long-term experiments are also available for other ecosystems. In ecological research (LTER), agriculture and grassland (Rothamsted Research), soil science (LTE), or agroecosystems (LTAE), long-term observations have similar importance (Blake 1999; Redman et al. 2004; Körschens 2006) as in forest ecology. Compared with other ecological fields, long-term experiments in forests are even under higher pressure due to their particular longevity and high space consumption. As forest land is often disregarded by politicians and other decision makers and extensively transformed into agricultural areas, urban building, and infrastructures, long-term experiments are frequently sacrificed and valuable long time series of survey discontinued.
It is an often heard, but misleading argument, that forest inventories that have been established during the last few decades at national or enterprise levels can substitute for the information potential of long-term experiments (Gadow 1999). Certainly, temporary plots of forest inventories may be harnessed by innovative Big Data methods, such as geospatial random forests and geostatistical mixed-effects models (e.g. Liang et al. 2016). For obtaining information about the status quo at the enterprise or landscape level, inventories are ideal, as this is what they are designed for. Furthermore, forest inventories provide the data for initializing simulation models, for forest planning and scenario analyses.
In the following, we summarize the main reasons why long-term experiments excel by far the information potential of temporary plots and why they are highly important for ecological monitoring and research (Pretzsch et al. 2019). First, only long-term experiments allow to study the long-term structure and growth development at the tree and stand level. They allow revelation of changing ecological processes, legacy and ecological memory effects, and the long-term consequences of ecosystem alteration. In contrast, temporarily observed inventory plots often lack information about the stand history and any intermediate yield before the plot establishment. This lack of information can be only partly remedied by establishing permanent inventory plots.
Second, long-term experiments are necessary for revelation of the cause-effect relationships between various stand treatment options and the response at the tree and stand level. In contrast to inventory plots, experiments usually analyse tree and stand reaction ceteris paribus, i.e. under parity of tree provenance, stand history, site conditions, and other internal and external factors. Temporary plots may indicate correlations but no evidence for casualties. In contrast to experiments, they may vary in many (even unobserved) traits and not only in the analysed factor of interest (Gamfeldt et al. 2013).
Third, long-term experiments usually cover un-thinned or untreated variants, which, for instance, indicate the maximum stand density, unfertilized conditions, etc. and serve as a reference for the density and growth of other treatments (e.g. thinned, mixed, fertilized plots). In addition, long-term experiments often comprise extreme variants that are usually avoided by forest practice and not covered by inventories. For analysing and modelling forest dynamics, however, extreme treatment variants are often even more informative than standard and business-as-usual variants mainly covered by inventories.
Fourth, only long-term surveys provide a full insight into the growth and yield of the remaining and removal stand, i.e. they quantify the total production since stand establishment. This includes also the otherwise not accessible information about the intermediate yield, caused by natural mortality or/and silvicultural treatments.
Fifth, long-term observations often cover a long part of the rotation or even include the subsequent stand generation. In this way, they can reveal effects of changes in environmental conditions at the tree and stand level (Spiecker et al. 1996). As they provide time series of growth and yield data that reach far back in time, they may provide evidence of environmental changes and the human footprint on forest ecosystems. The argument, which similarly long time series can be always obtained by retrospective growth-ring analysis, is only partly true. Retrospective analyses are possible at the individual tree level, but not at all for whole forest stands. This is because only the trees living at the time of sampling can be analysed retrospectively, but not their neighbours and the part of the population, eliminated before due to self-thinning or silvicultural interventions decades before. However, the past development of the population may be important to know its future dynamics (Pretzsch et al. 2021).
5.13.3 Need for Further Coordination and Standardization of Experimental Design and Set-ups
The foundation of the International Union of Forest Research Organizations (IUFRO) in 1892 was essential for the coordination and standardization of experimental design and set-ups. The recommendations and definitions for establishment and steering of thinning trials in the early times of IUFRO or its precursor organizations (Verein Deutscher Forstlicher Versuchsanstalten 1873, 1902) had some long-term standardization effects on many kinds of experiments in forests (Becking 1953). However, the objectives and questions of experiments changed and so did the variables and methods to measure. Among others, the measurements were extended to spatial explicit information about coordinates, crown size, stem quality, and vitality at the individual tree level. Natural regeneration, canopy characteristics, as well as variables quantifying the biodiversity, protective functions, recreational, or climate smartness are either added to the variable set of existing experiments. Or they are on the protocol list of new experiments from the beginning on.
Standardizing how to consider mixing proportions would be helpful, as they determine the establishment, steering, and evaluation of experiments (Dirnberger et al. 2017; Halofsky et al. 2018). Standards for height curve, form factor, or diameter growth-diameter functions would harmonize the evaluation of experiments and comparability of their results. Standardization of the result variables of long-term experiments, e.g. regarding merchantable volume, stem volume, stem mass, or total mass, would simplify the common overarching evaluations as realized in some of the studies underlying the book in hand (see Sects. 5.5, 5.6, and 5.7 and Hilmers et al. 2019; Pretzsch et al. 2020a, b; Torresan et al. 2020).
When establishing new experiments, addressing tree species mixing, natural regeneration, or transitioning from age-class forests to continuous cover forestry requires new standards for quantifying and steering mixing portions, stand density, and canopy cover over regeneration or for the horizontal and vertical stand structuring (del Río et al. 2016).
The date of measurement within the year, the frequency and methodology of measurement, and qualitative assessment of stem quality, crown vitality, etc. need some standardization, necessary for later common evaluation.
The silvicultural prescriptions for steering the experiments need common standards, analogous to the early thinning experiments, but more sophisticated as complex forest requires more detailed protocols for quantitative and reproducible, objective experimental steering (kind, strength, and frequency of interferences).
The common standards for evaluation at the stand and tree level simplify the pooling of data. A report of unique essential results will hopefully advance and support the appreciation and support of long-term experiments due to their unique contribution to forest observation, monitoring, and stewardship.
The standard for data storage and exchange will further support this ambition and save a lot of processing, organization, formatting, and time on the long term.
Trans-geographical projects with their international partner groups have the potential to further update the standard for experiments and observation in forest ecosystems from their establishment to the data storage and common evaluations.
5.13.4 Maintenance of Both Unmanaged and Managed Observation Plots
Untreated plots are of special value as reference for natural ecosystem dynamics without direct interference but suitable for revelation of indirect anthropogenic effects. They may reveal and quantify the effects of acid rain (Spiecker et al. 1996), climate change (Pretzsch et al. 2014), and also just local disturbances caused by lowering the groundwater level (Pretzsch and Kölbel 1988) or ozone (Matyssek et al. 2010). The growth and structure in managed forests is often superimposed by management effects in a way that disturbance by management and environmental stress are difficult to separate.
Unmanaged does not mean unmeasured; untreated plots are often accurately measured regarding vitality, growth, mortality, standing, and lying deadwood. Already the early thinning experiments provided un-thinned plots as reference; the inventory of all dropout trees provided information about the total yield over the whole rotation, not available without such concepts.
The reference plots offer information about maximum stand density, self-thinning, natural intra- and interspecific competition processes, and disturbances. They provide information for the derivation of basic relationships (Assmann 2013), model
parameterization (Pretzsch et al. 2002), and development of silvicultural prescriptions for both monospecific and especially mixed-species stands (Kelty 1992; Dieler et al. 2017). Any treatment experiments gain in value if untreated plots are nearby and demonstrate the impact of treatment by visual and quantitative comparison of both. In this network, we used the untreated plots for exploring trends in stand and tree growth that might be superimposed by various kinds of management effects on specifically treated experimental plots or un-specifically managed inventory plots (von Gadow and Kotze 2014). Plots with trees growing solitarily would be of similar interest but are even rarer (Kuehne et al. 2013; Uhl et al. 2015). They would reflect the vitality, growth, allometry, and ageing without competition, which would be of similar interest (for understanding, model
parameterization, biomonitoring) as growth under self-thinning and without treatment.
5.13.5 The Relevance and Perspectives of Common Platforms for Forest Research
The use of both long-term experiments as well as temporary plots with increment coring for detecting growth trends reveals the shortcomings of temporal plots for reliable information of growth trends. There are shortcomings of using temporary plots for tree-level evaluations as well as using them for retrospective stand-level growth trend diagnosis.
Trend statements about tree-level growth based on increment cores from temporary plots may be misleading if the tree history is unknown. Sampling in mature stands may lead to biased results, as it means a sampling of survivors that may not represent the mean growth of the population. Notice that a European beech stand may start with a million trees or so and arrive at 50–100 trees per hectare in the mature phase with a mortality rate of 1–7%. The dropout trees may be less vital and lower in growth, and the growth of the remaining trees is higher than the mean (Nehrbass-Ahles et al. 2014). The effects of unknown silvicultural treatment in the past, natural suppression in the understorey, or biotic and abiotic damages (e.g. browsing, acid rain) may have influenced the life history and still have a memory effects of the present and future growth of the trees. By selecting sample trees with long and large crowns indicating permanent dominance and with tree-ring patterns without narrow ring phases by suppression may reduce such biases but cannot completely avoid them.
By sampling on long-term experiments or observation plots, in contrast, it is possible to select representative trees without prior charge by suppression, silvicultural impact, or damages that may interfere with the normally sigmoid growth-age relation.
For the detection of any stand growth trend, information about the growth of trees of various social positions in the stand and about tree removal, mortality, and stand structure in the past is even more important (Torresan et al. 2020). In this platform, we tried to avoid misjudgements by selection of stands un-thinned at least in the near past, by sampling trees for retrospective growth calculation from all biosocial classes over the whole stem diameter range and by assessing tree mortality via stump inventory. In this way, stand growth can be derived retrospectively (Heym et al. 2017, 2018).
All these shortcomings underline the advantages of maintaining a network of long-term experiments for monitoring the vitality and growth of forests (Pretzsch et al. 2019) that after all cover one third of the land surface in Europe and deserve special stewardship, especially in the mountain
areas with their manifold ecological, economical, and socio-economic functions and services (Biber et al. 2015; Dieler et al. 2017).