One can study terrestrial ecosystems in many ways: as an attentive observer, as an experimentalist designing hypothesis-driven experiments, or as a modeller. Depending on the research questions asked and where the focus of this system-oriented and often also process-oriented research is, one not only carries out measurements of biogeochemical pools and fluxes but also quantifies responses of individual species and plant communities to changes in environmental conditions, plant diversity or ecosystem management. Studies can be short-term or long-term; they can be carried out at a single site or can rely on multiple sites. However, all of these different approaches, measurements and/or models need to account for the ecosystem characteristics mentioned in Chap. 13. Thus, the methodology used to study terrestrial ecosystems is often based on many different disciplines (e.g. biology, soil science, hydrology and micrometeorology) and is typically used in combination (e.g. measurements of biospheric–atmospheric gas exchange of entire ecosystems). This leads to interdisciplinary research, which demands solid disciplinary background knowledge but also interest and skills in interacting with neighbouring disciplines.

Which measurements are taken when, where and how often depends on the research question:

If one is interested in the effect of drought on water use in a grassland, then different water supplies to this grassland need to be taken into account—that is, precipitation, soil water storage, groundwater and maybe even irrigation. Thus, measurements of climatic variables such as precipitation and temperature, but also soil moisture, evapotranspiration and stand biomass and architecture, will have to be taken close to the field site (climate) and at the field site (all other variables), including soil profile measurements down to a certain depth, either continuously or in campaign mode before, during and after the drought.
However, if one is interested in the effects of drought on competition for water in a grassland, then the vegetation composition needs to be taken into account as well (Chaps. 19 and 20), in addition to the variables mentioned above. Thus, measurements of growth and ecophysiology (Chaps. 9 and 10) will have to be added at species or functional group levels and carried out in campaign mode in the field or in the greenhouse.

Overall, good knowledge of experimental design and statistics is crucial for the study of ecosystems, whether one focuses on observations or experiments.

Furthermore, long-term studies are necessary to separate short-term responses of ecosystems to environmental factors from long-term trends. If one is interested in the impacts of global warming on forest productivity, then short-term responses to a summer heatwave causing water stress in a particular year might counteract a long-term trend of increased productivity. Long time series of multiple decades are needed to separate these short-term signals (sometimes also called noise) from the long-term signal of global warming, particularly when the driver (air temperature) increases only slowly. However, such long-term data are rarely available. On the other hand, short-term studies, often carried out as campaigns of several days to weeks or as 3-year experiments (the typical funding period), need to take transient effects into account. Such effects occur when environmental conditions have been changed very suddenly in an experiment (e.g. by changing fertilisation or by establishing a new species composition) while the processes under study (e.g. plant growth or soil microbial processes) respond much more slowly in comparison with the change in the driver. For example, plant growth will be related not only to the new fertilisation regime but also to the large pool of nutrients already present in the soil. Thus, changes in plant growth in response to the new fertilisation regime might not become apparent until much later, when the old pool of soil nutrients is really affected by the new fertilisation. Using chronosequences (i.e. multiple sites subjected to the same driver but at different times) might be the way to go. For the example discussed above, a chronosequence with multiple sites (typically unreplicated) at which fertilistion was changed 1, 3, 6, 9 and 12 years ago might be better suited to study the responses to a change in the fertilisation regime. However, replacing time for space is based on the assumption that conditions were constant over time, which is not always the case (Sect. 14.1.2 for further details).

Spatially distributed studies are necessary for generalisation of the findings from otherwise individual case studies. Thus, single sites are part of gradients (typically along environmental drivers or management intensities) or transects (typically across the landscape or across continents), networks or grid-based inventories (Fig. 14.1). Definition of the variables and criteria on which these spatially distributed studies are based is relatively easy (e.g. annual temperature, across Australia, grid size 500 × 500 km). However, quantification of the abiotic and biotic factors that covary with these criteria (e.g. annual precipitation, soil type, elevation and management), and identifying these interactions, is very difficult and thus often limits the applicability of such spatial approaches (Sect. 14.1.3).

/epubstore/S/E-D-Schulze/Plant-Ecology/OEBPS/images/72100_2_En_14_Chapter/72100_2_En_14_Fig1_HTML.png — Fig. 14.1
Different approaches to study terrestrial ecosystems. The approaches described in this chapter differ in complexity and possibility to control for environmental conditions, as well as in their spatial applications

14.1 Observations

Observations are often used to investigate processes and patterns in terrestrial ecosystems; that is, no additional manipulations of the natural conditions are carried out, but the actual environmental conditions the ecosystem is experiencing are studied. Typical research questions addressed with this type of approach are: How does forest productivity react to changes in climate? How does the native species composition change in response to a non-native organism invading the system? Sometimes observations are distinguished from monitoring. Although this distinction is not very well defined, one could argue that measurements done for a legal purpose—for example, for pollution control or for habitat assessment required for national or international treaties—become a monitoring effort. In contrast, measurements driven by scientific objectives are often called observations.

Nevertheless, ecosystems under natural conditions can also be subjected to “treatments” or “experiments”. Comparison of different management intensities of agricultural systems might qualify, for some researchers, as comparison of different treatment levels of an experiment. Climate change has been called an “inadvertent global experiment” affecting the Earth system in an unprecedented way (Ramanathan 1988). Thus, the study of ecosystems over very long time periods might allow assessments of their responses to global change or to natural hazards. Ideally, such time series should be replicated at different locations, although unique time series exist for single locations (e.g. the CO₂ concentration record at Mauna Loa, Hawaii, USA).

14.1.1 Whole-Ecosystem Observations

Whole-ecosystem observations started with the onset of plant geography, when scientists and explorers such as Willdenow, von Humboldt, de Candolle and Grisebach described vegetation composition and plant distribution around the world in the eighteenth and nineteenth centuries. Later, experimental studies were added by Stahl, Kerner von Marilaun, Warming and Schimper, along with physiological, histological and climatological aspects in the twentieth century (Buchmann 2002; Sect. 14.2). Whole-ecosystem observations gained new momentum with the International Biological Program (IBP; 1964–1974) from the United Nations Educational, Scientific and Cultural Organisation (UNESCO), originating in Europe and inspired by the major scientific success of the International Geophysical Year (IGY; 1957–1958). The IBP was followed by the Man and the Biosphere Programme (MAB; still running), launched in 1971 also by UNESCO, extending the scope of the science to include humans. Ecosystem ecology entered a new era, also called “big ecology” (Coleman 2010). Large-scale studies were complemented by regional to global networks, employment of new technology and new approaches to access, for example, tall canopies allowed completely new research questions to be addressed. For example, the World Climate Research Programme (WCRP) was established in 1980 to develop our scientific understanding of the physical climate system. Many studies on global environmental change were brought together within the International Geosphere–Biosphere Programme (IGBP), launched in 1986. In parallel, the International Human Dimensions Programme (IHDP) was initiated in 1990 to address the human and societal aspects of global change, and Diversitas was established in 1991 to study the loss and change of global biodiversity (Greenaway 1996). All four programmes were brought together in 2001 under the umbrella of the Earth System Science Partnership (ESSP) to foster better cooperation among them and to support greater integration across disciplines, particularly between natural and social sciences. All of these efforts continued until the end of 2015 and have now been brought together in Future Earth (http://www.futureearth.org) to provide scientific knowledge to support the transition of global societies towards global sustainability (Part V of this book).

14.1.1.1 Large-Scale Case Studies

Many large-scale observational studies of ecosystems had their origin during the IBP, such as the German Solling Project (Ellenberg 1971) or the Grassland Biome Studies (French 1979). In the Solling Project, questions about the effects of acid rain on soils and forest ecosystems were the main focus, later followed by experimental manipulations of incoming precipitation. The time series of soil nutrient fluxes started at this time (Brumme and Khanna 2009), in its duration comparable to the atmospheric CO₂ concentration measurements at Mauna Loa. Some of the most prominent achievements were recognition of the relevance of below-ground processes for nutrient cycling, identification of atmospheric deposition as a nutrient source for vegetation, greater understanding of the year-round activity in/of ecosystems and confirmation of the usefulness of models to describe ecosystem processes. The Ulrich canopy budget model (1983) is a prime example of interdisciplinary work on the ecosystem scale (Chap. 13, Sect. 13.6), allowing interception deposition and canopy exchange to be estimated on the basis of meteorological input variables, physical surface characteristics, leaf area and stand structure information. At about the same time, the Hubbard Brook Ecosystem Study (HBES) started (Sect. 14.2), investigating the effects of forests on watershed hydrology and flood control, later followed by whole-watershed biogeochemistry and modelling. All of these ecosystem studies relied on measurements of many different components of and actors in a single ecosystem (e.g. soil, water, microorganisms, vegetation, fauna and atmosphere; Sect. 13.2), always carried out by many different research groups working at the same site and often embedded in national or international networks.

14.1.1.2 Regional to Global Networks

Whole-ecosystem observational studies started out as single-site studies, sometimes already with the intention to be run long-term, but it became obvious that such studies were inherently limited in terms of spatial representativeness and replication (Chap. 13). Thus, the scientific community realised during the IBP in the early 1970s that networks of sites help to overcome this limitation, providing the necessary statistical rigour and a wide range of further benefits: exchange of ideas and methodologies; upscaling to larger regions, biomes, continents or even the globe; and identification of global trends in environmental conditions and the respective response of the terrestrial biosphere, to name just a few. Thus, national networks evolved, such as the US Long-Term Ecological Research network (LTER; since 1980; http://www.lternet.edu/), which in 1993 was expanded into the International Long-Term Ecological Research network (ILTER; https://www.ilter.network//), a global network of networks aiming to understand the response of ecosystems to global change. Similar networks were created within the framework of international conventions, such as the International Co-operative Programme on Assessment and Monitoring of Air Pollution Effects on Forests (ICP Forests; http://icp-forests.net/), monitoring forest conditions at 6000 plots in 42 countries in 2016. Taking on monitoring tasks and also responding to the increasing demand for easier access to and greater usability of data, the European research infrastructure project Integrated Carbon Observation System (ICOS RI; https://www.icos-ri.eu) was created in 2015. This pan-European network of long-term measurement stations for greenhouse gas emissions and their regional dynamics includes not only terrestrial but also atmospheric and marine stations. Within these networks, a high level of standardised sensor specifications (e.g. sensor drift due to sensor ageing or ambient temperatures), harmonised protocols (e.g. sampling details and calibrations) and experimental set-ups (e.g. placement of measurement devices) enable comparable measurements to be taken. This in turn is a fundamental prerequisite for spatial and temporal upscaling for reliable identification of ecosystem responses to environmental change.

14.1.1.3 New Technologies

With the development of instruments to measure CO₂ and H₂O vapour exchange at very high temporal resolution (at 20 Hz, i.e. 20 times per second), a new type of ecosystem-scale study became possible in the early 1990s. By use of these new instruments for the so-called eddy covariance (EC) method, ecosystem CO₂ and H₂O vapour exchange with the atmosphere could be quantified. The EC method is based on measurements of a turbulent flux in the air above an ecosystem (the measurement height is about one third higher than the stand height). The vertical wind speed and the gas concentrations in air parcels moving past the respective sensors are measured with high temporal resolution (among other meteorological variables). Assuming that advection is negligible, the covariance of the vertical wind speed and gas concentration (i.e. the product of the deviations of both variables from the mean, typically over 30 min) is a direct measure of the net flux. The ability to measure the biospheric–atmospheric gas exchange of an ecosystem continuously—that is, 24 h a day, 7 days a week, 365 days a year—for several years, instead of upscaling manual chamber measurements of soil respiration and evaporation combined with leaf-level gas exchange taken during weekly or monthly campaigns, suddenly allowed a much deeper insight into temporal changes of ecosystem processes than ever before. On the basis of micrometeorological theory (Aubinet et al. 2012), the physical origin of whole-ecosystem fluxes could be determined (e.g. footprint analysis) and fluxes could be partitioned into component fluxes, such as gross primary production (GPP; i.e. CO₂ uptake) and total ecosystem respiration. Nowadays, Fluxnet (http://fluxnet.fluxdata.org/about/), a global network of micrometeorological flux sites, consists of >850 flux tower sites located on five continents, providing more than 6600 site-years of measurements (Fig. 14.2).

/epubstore/S/E-D-Schulze/Plant-Ecology/OEBPS/images/72100_2_En_14_Chapter/72100_2_En_14_Fig2_HTML.png — Fig. 14.2
Flux sites where biospheric–atmospheric gas exchange is measured using the eddy covariance technique. a Global map of Fluxnet, showing the locations of sites as of October 2015 (image courtesy of D. Baldocchi; http://fluxnet.fluxdata.org/sites/site-summary/). b 35 m tower at the Davos evergreen forest site (Graubünden, Switzerland). The instrumentation for eddy covariance is installed on the top beams visible above to highest platform (Photo courtesy of L. Hörtnagl). c 2 m mast at the Chamau lowland meadow site (Zug, Switzerland) (Photo courtesy of L. Merbold). Both sites are part of the Swiss Fluxnet, the network of flux sites in Switzerland (http://www.gl.ethz.ch/research/bage/fluxnet-ch.html)

On the basis of such measurements, carbon sequestration of terrestrial ecosystems can be directly estimated by summing up fluxes over a growing season or 1 year. It has turned out that forests, after reaching canopy closure, are typically a carbon sink, even at very great age (several hundreds of years), while very young and open forests are typically a carbon source (despite the trees growing) due to high respiratory CO₂ losses from the soil (Buchmann and Schulze 1999; Kolari et al. 2004; Magnani et al. 2007). Thus, these flux measurements challenge the long-standing theory of Odum (1971) that old forests are at carbon equilibrium (i.e. with a ratio of gross production to community respiration approaching 1) and thus respire almost as much as they assimilate, resulting in a carbon budget of zero at mature stand stages, being neither a sink nor a source. However, this theory, prominently presented in Odum’s textbooks Fundamental Ecology (Odum 1971, 2005) and Basic Ecology (Odum 1983), was based on a hypothetical model of a 100-year age series of dense forest stands by Kira and Shidei (1967), who rather focused on the fact that net forest productivity is the balance between gross production and respiration. Ecosystem flux measurements with this new technology clearly proved that old forests are not at carbon equilibrium. Further technological progress (e.g. the newest developments in laser spectrometry) nowadays allows not only measurement of concentrations of trace gases such as CH₄ and N₂O but also measurement of stable isotopic signatures of all highly relevant gases such as CO₂, H₂O_vapour, N₂O and CH₄, at high temporal resolution (e.g. 10 Hz for CH₄ and N₂O; 1–5 Hz for ¹³C and ¹⁸O in CO₂). This now offers the unique opportunity to investigate the biological origin of these molecules (Wolf et al. 2015).

14.1.1.4 New Approaches to Access Tall Canopies

Not only networks of sites but also the development of new approaches to access and to observe previously inaccessible compartments of terrestrial ecosystems have enabled new insights into ecosystems. These new approaches to forest canopies are:

Canopy rafts linked to blimps—the “Radeau de Cimes”, first employed in French Guiana in 1986 (Hallé et al. 2000), then later on also in Cameroon, Gabon and Panama.
Canopy walkways—first employed in Hopkins Forest (MA, USA) in 1991, then refined and replicated globally (Lowman 1999).
Canopy cranes—first employed in Panama in 1990, then replicated globally and also organised within a network—the International Canopy Crane Network (ICAN) (Stork et al. 1997; Sutton 2001).

All of these have allowed access to 30 to 50 m tall canopies, formerly accessible only via tree climbing. New insights into biological diversity present in tropical tree canopies, canopy structure and plant ecophysiology were the results. Currently, a new tool is emerging: the rapidly increasing employment of drones (or unmanned airborne (or aerial) vehicles (UAVs)). They can be used for observation of ecosystems from above, including spatial assessment of species composition, foliage phenology and biochemistry, as well as stand structure and spatial extent of certain systems or disturbances, depending on the sensor type flown (Anderson and Gaston 2013). Thus, drones can close the gap to remote sensing techniques (Sect. 14.1.4). UAVs also allow revisitation of field sites at frequent intervals and flights at low altitudes for fine spatial resolution, at relatively low operating costs, but special permits might be needed.

14.1.1.5 Observations of Biodiversity

Questions as to how biodiversity affects ecosystem functioning has led to various approaches being used. Certainly the first approach that comes to mind is observation of natural communities that differ in one aspect of biodiversity (e.g. the number of species present or the abundance of a certain functional group; Chap. 20) and comparison of them in terms of a variety of ecosystem processes. Thus, several forest stands would be sampled with a random or grid-based selection of plots across a specific area, ecosystem functions (e.g. biomass production) measured and then related to the observed biodiversity (e.g. the number of tree species). Such monitoring or sample surveys reflect natural conditions with respect to the age structure, canopy and root architecture, food webs or biogeochemical cycles of well-established sites (Leuschner et al. 2009), thus, transient effects do not occur. In addition, such an approach capitalises on the large number of permanent forest inventory plots existing in many countries (national inventories, ICP Forests; see above). However, usually, not every level of biodiversity is equally represented; often plots have biodiversity levels close to the mean or even close to the low end of diversity in managed forest ecosystems (Vilà et al. 2007). In addition, unless site conditions are almost identical, such cross-habitat or cross-site comparisons may hide effects that biodiversity exhibits within a site, because environmental differences between sites introduce “noise” into the diversity–function relationship (Sect. 14.1.2). Moreover, these environmental covariables might determine the diversity of an ecosystem and hence ecosystem properties and processes, and plotting biodiversity orthogonally to ecosystem function has thus been considered a “strange thing to do” (Naeem et al. 2009). For example, an asymptotic increase in biomass production with increasing tree species richness could be due to functional differences among species, leading to niche differentiation, higher resource exploitation and hence higher productivity. However, more productive stands may simply permit the coexistence of more species (Chap. 20). Thus, monitoring or sample surveys can be used to document correlations between biodiversity and ecological processes; however, causal relationships can only be approximated by accounting for measured (!) environmental covariables, using suitable statistical procedures such as structural equation modelling (e.g. path analyses). To avoid some of these confounding effects, comparative studies have been designed, where a similar number of plots per diversity level are deliberately chosen along gradients of biodiversity, ideally coupled with maximum standardisation of environmental conditions, including stand age or management history (e.g. Baeten et al. 2013). Nevertheless, there are many environmental variables influencing biodiversity, so the selection of study sites remains subjective.

14.1.2 Transects and Chronosequences

Transects (i.e. a collection of sites along an imaginary line, most often along an environmental gradient) and chronosequences (i.e. a collection of sites subjected to the same driver but at different times) are additional approaches used to address research questions in ecosystem ecology. Typical research questions are: How does increasing precipitation/temperature/soil fertility affect stand productivity? How does carbon allocation above-ground versus below-ground change with increasing altitude? Does soil carbon sequestration reach saturation eventually? What are the effects of fire on species composition?

14.1.2.1 Transects

Assembly of a transect requires a strong gradient in the environment (e.g. climate, soil fertility, elevation) while other potential environmental drivers of ecosystem processes are kept either relatively constant or can be measured. For example, the elevational and thus the climatic transect on the island of Hawaii offers unique conditions. Ranging from sea level up to >2400 m in elevation the precipitation gradient ranges from 500 to >5800 mm annual precipitation and the air temperature gradient ranges from about 10 to >20 °C mean annual temperature. At the same time, all sites are located on the same bedrock—that is, recent volcanic substrates (Vitousek et al. 1992; Allison and Vitousek 2004). Along these environmental gradients, nutrient cycling, decomposition and vegetation composition have been shown to be intricately linked. Even invasion by six exotic invasive species has been shown to be related to the nutrient status of the native vegetation.

Many very long transects (>1000 km in length) were established in the early 1990s by the international Global Change and Terrestrial Ecosystems (GCTE) core project within the IGBP to study large-scale responses to global change (Walker et al. 1999; Canadell et al. 2002). These long transects have proven ideal to provide data for large-scale synthesis and integration efforts and thus the necessary large-scale information for global models. Some transects have been used over more than a decade, such as the sub-Saharan and Kalahari transects, ranging from the arid subtropics to the moist tropics (Shugart et al. 2004), while others have been used less frequently, such as the Patagonian transect (Schulze et al. 1996). There also exist networks of transects—for example, the Australian Transect Network (ATN), which consists of seven transects spanning across different biomes and across large rainfall, temperature and land use gradients from the coast to inland areas (http://www.tern.org.au/). Overall, the lengths of transects vary substantially, ranging from several thousand kilometres (such as in sub-Saharan Africa) to less than 100 m in Sweden (Högberg 2001). This illustrates that environmental gradients can act on very different scales, across individual slopes to across continents, creating a mosaic of ecosystems across the landscape.

14.1.2.2 Chronosequences

Chronosequences are often used when the impacts of drivers are studied that trigger rather slow reactions of the ecosystem over long time periods (e.g. responses to slow continuous forcing) or when transient responses of ecosystems are expected (e.g. responses to sudden disturbances; Sect. 14.1.5 and Chap. 17). Preferentially, only the time since the driver changed should differ between the sites forming a chronosequence. In reality, this is extremely difficult to achieve, since other state variables might have changed over time as well—that is, variables that describe the ecosystem well enough to predict its future behaviour in the absence of any external driver affecting the system (Jenny 1941). Such state variables can be climate, soil properties, previous management practices, vegetation composition or microbial communities. If changes in state variables are expected, in addition to the main driver studied, the experimental set-up and measurement portfolio need to be adjusted accordingly—that is, by adding actual measurements of these variables.

Studying the long-term response of tree growth to atmospheric nitrogen (N) deposition can be done only using a chronosequence approach, since experimental manipulations would simply take too long. While spruce trees grew until the 1940s according to well-established growth isolines based on site quality and yield tables (Fig. 14.3), stem growth rates increased strongly between 1940 and 1970. In turn, trees of all age classes “crossed” growth isolines, concomitantly with increased atmospheric N deposition, thus growing better than traditional forestry knowledge (collected before the 1960s) would have suggested. However, between the 1970s and the mid-1980s, growth rates ceased to increase, at a time when strong forest decline symptoms were observed throughout central Europe, linked to high S and N deposition. Since the late 1980s, growth rates have increased again, related to a decrease in atmospheric S pollution but continued high N deposition (Fig. 14.3) (Mund et al. 2002).

/epubstore/S/E-D-Schulze/Plant-Ecology/OEBPS/images/72100_2_En_14_Chapter/72100_2_En_14_Fig3_HTML.png — Fig. 14.3
Long-term growth patterns of spruce trees. a Average annual stem volume increment of trees growing in different aged spruce stands (*Picea abies*) over 142 years. b Annual stem volume increment as a function of tree age. (Mund et al. 2002)

Although this example nicely illustrates the power of a chronosequence approach, this study relied on certain assumptions regarding other state variables, as discussed above: all trees should have had the same genetic origin, and all stands should have had comparable soil and climate conditions, as well as comparable management practices (e.g. thinning). Most of these assumptions were explicitly tested in this study; thus, the general interpretation is rather robust.

Overall, both approaches have their advantages and disadvantages; thus, awareness of their shortcomings is a necessary first step to avoid erroneous data interpretation. Nevertheless, use of transects and chronosequences has resulted in unprecedented and detailed insights into spatial patterns and ecosystem adaptations across space and time.

14.1.3 Grid-Based Inventories

Grid-based inventories are often used in monitoring studies to increase spatial representativeness, to avoid experimental bias (e.g. due to subjective plot selection) or to provide inputs into geographical information systems (GIS). Typical research questions are: How severe is the effect of insect outbreaks/fire/flooding across the country? Is there a relationship between species diversity across a landscape and land cover fragmentation? How has soil fertility management affected agricultural productivity across a region?

The location where information is available or measurements are taken is determined on a regular grid. Grid sizes can vary depending on the research questions and data availability, from small grid sizes of 0.5 × 0.5 m for small field studies (e.g. on vegetation cover or soil properties) to medium grid sizes of between 500 × 500 m and 10 × 10 km in regional or national studies to large grid sizes of 0.5°× 0.5° in global model runs. Aggregation (or disaggregation) questions come into play when data are available at different grid sizes but need to be combined. Spatial uncertainty needs to be considered if the data are valid for scales smaller than the final grid size used—for example, by calculating error correlation lengths or semi-variograms (see spatial statistics textbooks for details). Often information about many different variables is gathered at the same grid size to overlay this information to gain new insights. For instance, to construct the first spatially explicit Swiss CH ₄ emission inventory (Hiller et al. 2014), information about the spatial distribution of livestock and agricultural farms, and also terrestrial ecosystems such as forests, grasslands, wetlands and lakes (and reservoirs) as well as population density, urban areas and waste facilities was used (Fig. 14.4).

/epubstore/S/E-D-Schulze/Plant-Ecology/OEBPS/images/72100_2_En_14_Chapter/72100_2_En_14_Fig4_HTML.jpg — Fig. 14.4
Role of grid sizes in spatial analyses. Spatially explicit CH₄ emission inventory at grid sizes of a 500 × 500 m and b 0.1° × 0.1° equivalent to 10 × 10 km (Hiller et al. 2014). The emissions depicted in panel a include both natural and anthropogenic sources (which made up >95% of total emissions), while panel b depicts anthropogenic CH₄ emissions according to the EDGAR v4.2 inventory only. Note the coarse resolution in panel b compared with that in panel a, reducing the information content tremendously

14.1.4 Remote Sensing

On even larger scales—that is, on continental or global scales—only remote sensing approaches can provide the necessary information on terrestrial ecosystems. Today, the term “remote sensing” typically refers to airborne sensors (e.g. on satellites, aircrafts, drones), which either emit a signal (active remote sensing) or simply record a signal (passive remote sensing). Often, multispectral platforms (i.e. multiple wavelengths of electromagnetic radiation) are used, such as Earth observation satellites (e.g. Landsat and IKONOS). Landsat data are available since the 1970s, with spatial resolutions between 15 and 100 m and pass-over times every 16 days, while IKONOS data are available since 1999, with spatial resolutions between <1 m and 3.2 m and pass-over times every 3 days. Typical research questions are: Where are the hotspots of land degradation/productivity/species diversity/deforestation across a continent? How do multispectral signals relate to plant and stand performance across large regions?

Depending on the research question, different remote sensing products can be obtained, often available at almost daily to biweekly intervals, sometimes even free of charge (e.g. Landsat, Moderate Resolution Imaging Spectroradiometer (MODIS) and Advanced Very High Resolution Radiometer (AVHRR)). Many of these products from passive remote sensing are based on the spectral properties of surfaces—that is, reflectance (or absorbance) of light at different wavelengths: from visible light (0.38–0.72 μm) to near-infrared (0.72–1.3 μm), short-wave infrared (1.3–3 μm), mid-wave infrared (3–5 μm) and long-wave infrared (8–15 μm). Note that the subdivisions for infrared light are not precise and depend on their applications. Reflecting surfaces can be those of plant tissues, soil, water or man-made structures. For plant foliage, the leaf biochemistry, leaf structure and foliage distribution within a canopy, but also the leaf colour and leaf water content, are important factors affecting the reflectance of light at different wavelengths (Chap. 9). While moist green foliage mainly absorbs blue and red light but reflects green light (because of chlorophyll) and shows low reflectance in near-infrared and short-wave infrared wavelengths, dry brown or senescent foliage reflects more in the red wavelengths (because of carotenes and other pigments) than in the green wavelengths and shows high reflectance in short-wave infrared wavelengths.

These differences in the reflectance of different wavelengths are then used to calculate vegetation indexes such as the Normalised Difference Vegetation Index (NDVI), the Photochemical Reflectance Index (PRI) and the Enhanced Vegetation Index (EVI), to name just a few important ones. While the NDVI scales positively with biomass and the leaf area index (LAI; until about an LAI of 6), the PRI provides seasonal information about the ratio of chlorophyll to carotenes. Both indexes have been successfully used to assess terrestrial productivity and its changes, particularly over continental areas or globally (Myneni et al. 1997; Running et al. 2004). The EVI is more sensitive to high biomass and thus high-LAI areas than the NDVI, also better reflecting stand architecture. More recently, sun-induced fluorescence (SIF) has been used to estimate GPP of terrestrial vegetation (Guanter et al. 2014). All of these indexes and proxies, when they relate to plant activities, can also be used to upscale point measurements to larger areas on the basis of data acquisition from airborne sensors (UAVs, aircrafts, satellites). However, such indexes need to be checked with adequate ground truthing before they can be used as a basis for modelling (Chap. 22).

Active remote sensing—that is, the employment of RADAR (RAdio Detection And Ranging) and LiDAR (Light Detection And Ranging)—is based on the time delay between the emission and return of a signal. RADAR is based on the emission of radio waves, which are reflected or scattered by objects, and is used for purposes such as large-scale digital elevation models. The advantage of RADAR (using radio waves) over LiDAR (using visible, infrared or ultraviolet light) is the low absorption of radio waves by the medium they pass through—for ecological applications, mainly the medium of air. Thus, clouds, fog or rain limit RADAR applications much less than LiDAR applications. For the use of airborne or terrestrial LiDAR, a laser beam is emitted, which is back-scattered by surfaces or molecules. On the basis of the recording of the return, the height and structure of vegetation or land surfaces can be determined/scanned. Coupling of both airborne and terrestrial LiDAR gives the best results.

Overall, remote sensing approaches have been widely used and have provided unprecedented insights. However, they also have drawbacks, and one needs to pay attention to their limitations. For example:

Remote sensing products are often proxies for a biological process or ecosystem characteristic. As such, they introduce uncertainty, since the proxy is never identical to the variable under study because their relationship is typically not 1:1.
Certain variables such as land use intensity (fertilisation levels, harvesting frequencies) cannot easily be seen from above or would need many images of the same scene.
Ground-truthing might still be necessary, particularly if new products are employed on scales smaller than the globe.
Depending on the approach chosen and the region under study, clouds might obscure the view from above. Then composite images are used—for example, over 10–30 days. One has to pay attention to whether this temporal resolution is then still adequate for the process under study.

14.1.5 “Natural Experiments”

Yet another approach to observation of terrestrial ecosystems is to use “natural experiments” or “accidental experiments”, such as sudden disturbances or slow continuous forcing (Chap. 13). Typical research questions are: How does a landscape recover after a large disturbance event? What is the fate of a pollutant within ecosystems and along food chains? What are the effects of global warming on ecosystem processes and the provisioning of ecosystem services?

Obviously, one cannot wait for such natural or man-made hazards to suddenly happen, but one can be attentive to make use of them once they do occur—for example, volcanic eruptions, strong storms, flooding, devastating insect outbreaks, large-scale fires or radioactive fallout/spills. Thus, long-term research efforts are needed, preferentially starting before but clearly after such sudden disturbances. Nevertheless, excellent examples are available—for example, the research following the Mt. St. Helens (WA, USA) outbreak in 1980, the large fires in Yellowstone National Park (YNP; USA) in 1988, the extensive bushfires in Australia between 2003 and 2009, and the devastating mountain pine beetle outbreak in North America in 2008. Sometimes research after such sudden disturbances leads to conceptual changes in science or even ecosystem management (e.g. fire management in YNP (Franke 2000; Barker 2005)) and might challenge long-standing theories (e.g. the diversity–stability relationship after flooding (Wright et al. 2015)).

These “natural” or “accidental” experiments have also been shown to provide great insight into ecosystem ecology, particularly when one is dealing with anthropogenic impacts on the environment—for example, when studying the effects of climate change or invasive species, or when determining the transport of particular substances in ecosystems (Table 14.1). However, such studies can also have drawbacks because information about the starting point of the “experiment” or defined boundary conditions might not be available (Diamond 1983; HilleRisLambers et al. 2013). Nevertheless, such “experiments” can be particularly important when scientifically planned experiments in ecosystems are not acceptable to the public (e.g. involving increased radioactivity or chemical concentrations in the environment after nuclear fallout or chemical spills, respectively) or when such experiments are not practicable (e.g. increasing atmospheric temperatures over Europe). Depending on the background information prior to the “accidental” experiment, valid conclusions might be drawn about the effects and ecosystem responses to the accidental driver. A further option to exploit “natural experiments” of slow continuous forcing—for example, acid atmospheric deposition, climate change or species invasions—is to combine observations with monitoring and dedicated experiments (Sect. 14.2). Then, also such slow changes in the environment can be used to learn about the mechanistic underpinning of ecosystem processes and species interactions (e.g. increasing frequency of drought events and tree mortality (Allen et al. 2010, 2015)), making a well-designed monitoring network even more important.

Table 14.1

Characteristics of different approaches to study ecosystems: observations and monitoring, experiments, and natural and accidental experiments. (Modified from HilleRisLambers et al. (2013))

Characteristics	Observations and monitoring	Experiments	Natural and accidental experiments
Control over natural perturbation	None	None to low (if counteracting measures are taken)	None
Control over treatment	None to low (land use)	High	Intermediate
Appropriate control	None to low (time series, chronosequence)	High	None to low (time series, chronosequence)
Spatial scale	Small to large	Typically small	Small to large
Temporal scale	Short-to long-term	Typically short-term	Short- to long-term
Difficulty of imposing treatments	Easy (no treatments)	Difficult	Easy (treatment already imposed)
Relative costs (investment, supplies, labour)	Low to intermediate	High (treatments have to be imposed)	Low to intermediate
Main advantages	Large spatio-temporal scales, ecosystem dynamics, high representativeness in networks, study of remote sites	Control over treatments, cause–effect relationships	Large spatio-temporal scales, opportunity for treatments not otherwise possible
Main disadvantages	Correlations and covariates limit cause–effect interpretation	Small spatio-temporal scales, treatment strengths subjective, treatments can fail, costs, logistics	Treatments unpredictable and infrequent, representativeness of treatment, correlations and covariates limit cause–effect interpretation

In general, all approaches described in Sect. 14.1 have their advantages and disadvantages, based on their diverse characteristics (Table 14.1). Whether there is control over naturally occurring perturbations in addition to potential treatment effects; whether a control exists (either a zero treatment or use of time series analysis or chronosequences); what spatio-temporal scales one can study; what difficulties and costs might arise—all of these aspects depend on the study approach that is taken. No one approach is better than the other; its selection is simply based on the research questions that are asked, as outlined above.

14.2 Experiments

When the study of terrestrial ecosystems under natural conditions is not sufficient to address the research questions asked or does not provide enough explanatory power, experiments need to be carried out in which the natural setting is altered (i.e. the treatment) and then compared with the natural condition, also called the ambient condition (i.e. the control). Experiments addressing ecosystem ecology are typically carried out in the field but are sometimes also done with “artificial” ecosystems under more controlled conditions (Sect. 14.2.5). A semi-artificial setting is the common garden approach, where plants from different locations are planted together in one location (i.e. the common garden) to study, for example, their plasticity in response to environmental factors.

14.2.1 Manipulations of Pools and Processes

Experiments focusing on processes, without changing the natural environmental setting, are often very close to observations. Treatments typically manipulate a selected pool (e.g. litter) or process (e.g. carbon allocation), and one observes this pool or process, as well as its interactions with other pools, processes and organisms in the ecosystem under study. Typical research questions are: Does soil respiration depend on canopy photosynthesis? How fast is the coupling between above-ground and below-ground? What determines litter decomposition under field conditions?

Such experiments are often carried out over one or multiple growing seasons, but not necessarily over decades or longer. The duration of such experiments is determined by the time needed to observe a response reliably and also to separate the treatment effect from intra- and inter-annual variations in the pool or process under study. Experiments sometimes include multiple levels or qualities of treatment. Litter decomposition studies, for instance, can be set up by changing the frequency (e.g. multiple harvest times of litter bags), the intensity (e.g. different litter amounts or mesh sizes in litter bags) or the quality (e.g. different litter types) of the treatment. They may also include multiple treatment combinations (e.g. litter quality × litter placement × litter amount). Thus, the experimental design becomes very important when one is setting up such uni- or multifactorial experiments, particularly in terms of (pseudo-)replication (for details, see statistics textbooks). It was on the basis of such litter decomposition studies that the relevance of the plant growth form was shown, which was even more important than direct climatic effects (Bardgett and Wardle 2010).

Experiments most often start with a clear research question (e.g. How are above-ground and below-ground processes linked?), a clearly formulated hypothesis (e.g. Carbon allocation below-ground is linked to canopy photosynthesis.) or an objective (e.g. to assess above-ground and below-ground interactions). Thus, the experimental treatment is aimed at manipulating the setting under which the process under study occurs (e.g. transfer of carbohydrates from the canopy to below-ground). The treatment can be designed to either (1) limit or prevent the transfer of carbohydrates (e.g. by girdling the tree stems—that is, cutting parts of the phloem away while leaving the xylem intact) when working in a forest (Högberg et al. 2001); or (2) to trace the freshly fixed carbohydrates below-ground (e.g. by labelling the carbohydrates with (stable or radioactive) carbon isotopes (¹³C or ¹⁴C, respectively)) by providing labelled CO₂ as a source for photosynthesis (forest: Högberg et al. 2008; grassland: Burri et al. 2014; artificial ecosystem: Ruehr et al. 2009). Measurements are taken according to the research question and the experimental set-up, and also need to assess the environmental conditions during which the experiment is taking place. In the example on allocation, ambient climate and soil conditions need to be measured (e.g. temperature, light, vapour pressure deficit (VPD), soil water content, soil texture) to explain any physiological process rate; above-ground and below-ground processes need to be assessed (e.g. photosynthesis, soil respiration, root exudation, microbial activity) to describe above-ground and below-ground activities; and samples need to be taken for isotope analyses (e.g. bulk tissues, foliage, phloem and root samples for carbohydrate extractions) to assess transfer rates of labelled photosynthates from CO₂ fixation in the canopy to arrival of labelled carbohydrates below-ground or loss of labelled CO₂ due to root or soil respiration. On the basis of such a measurement portfolio, soil respiration in a forest was shown to be clearly coupled to canopy photosynthesis, with about a 54% contribution of root respiration to total respiration during the growing season (Högberg et al. 2001). On the other hand, above-ground coupling slowed down under drought conditions in grassland (Burri et al. 2014) and artificial ecosystems with beech saplings (Ruehr et al. 2009). Thus, the duration of the experiment needs to be adjusted to plant and soil activities (e.g. the photosynthesis rate as the source process; and root or soil respiration, carbohydrate storage in roots and root growth rate as the sink processes) and also to canopy height, since transfer rates scale with canopy height (Kuzyakov and Gavrichkova 2006). Such an allocation experiment in a forest might take days to weeks to accomplish, but only hours to days in a grassland. The number of replicates (here, replicated plots) needs to be determined as a function of biological variability and also of time and financial constraints. Because of the latter, the average number of replicates is often three to five per treatment level or control in field ecology, although greater numbers are desirable.

14.2.2 Manipulations of Environmental Conditions

Experiments manipulating environmental conditions for entire ecosystems can be—and have been—done in various ways. One can manipulate the resources for terrestrial ecosystems, the climatic conditions under which all processes take place, the biogeochemical cycling (Sect. 14.2.1), as well as the organisms present in an ecosystem (Sect. 14.2.3). Focusing on the environmental factors, such manipulation experiments have a long tradition in ecosystem ecology to answer the following research questions: What are the effects of drought/high CO₂ concentrations/fertilisation on water and nutrient uptake/turnover in a given ecosystem? Do different species react differently to changing environmental conditions? How resistant/resilient is a given ecosystem to changes in climate? Thus, the following manipulations have been carried out:

Resources (i.e. water, nutrients or light): Drought and irrigation; N/atmospheric deposition exclusion and addition; elevated CO₂ or O₃ concentrations; liming, fertilisation; shading; root trenching.
Climate: Soil warming with heating cables, ecosystem heating with IR lamps, increased ultraviolet (UV)-B.
Various combinations of the above.

14.2.2.1 Transplant Experiments and Space-for-Time Experiments

One can set up quite an elaborate infrastructure to manipulate environmental conditions in a given ecosystem (see below) or use natural environmental gradients and move entire ecosystems into a new environment. Such transplant experiments thus use natural spatial variations, such as elevational gradients in climate, to study the response of ecosystems to changes in climate over time (space-for-time substitution). Transplanting alpine plants to lower elevations in the Swiss Alps, Alexander et al. (2015) were able to show the strong impacts of changing competitor identities on target plants’ performance: competitive pressure at the warmer, new site was more detrimental than the related climatic benefits. Likewise, gradients in soil development and fertility in glacier fore-fields can be used to study the temporal development and succession of emerging high alpine ecosystems. Transplant experiments obviously work only with small-statured ecosystems such as grasslands and/or small units of “ecosystems” (e.g. small monoliths of soil plus vegetation), which are transferred to the new location with great care (using means ranging from human power to helicopters) to avoid disturbances of the “ecosystem” (Ineson et al. 1998; Bassin et al. 2007). Such an approach has clear advantages, such as relatively easy logistics, fast set-up, small expense, and no need for manipulation infrastructure. However, this approach also bears great risks—such as disturbance of the soil profile, damage to root systems, large edge effects and different trophic interactions at the new location—and cannot be done for ecosystems with tall vegetation. The risks need to be known and either controlled or accounted for in the statistical analyses of the data. In addition, some research questions simply cannot be asked—for example, questions about trophic interactions or pest and pathogen pressures under the “future” climate (i.e. transplantation of vegetation to lower elevations where it is warmer), since the higher trophic levels, pests and pathogens have not been transplanted as well. The choice of treatments is also restricted by the existence of natural gradients, since environmental conditions must match the scientific objectives, as discussed by Beier et al. (2012).

14.2.2.2 Fertilisation and Irrigation Trials

One approach to manipulate the environmental conditions of whole ecosystems is to use experimental trials to increase resource supply. Originally employed in agricultural sciences (Sect. 14.2.4), the approach of fertilisation trials was adopted in the forest sciences only in the 1970s, often focusing on the expectedly most limiting resource, usually nitrogen or water. For example, trials on 30 × 30 m² Pinus sylvestris plots were established in northern Sweden by Olaf Tamm and co-workers, supplying the boreal forest with 34, 68 or 120 kg N ha⁻¹ year⁻¹ (compared with a low atmospheric background N deposition of about 3 kg N ha⁻¹ year⁻¹). These plots were followed over 30 years and yielded valuable insights into dose-related responses of tree growth to N additions, and also into tree recovery when treatments were stopped (Högberg et al. 2006). The larger the N additions were, the smaller the increase in tree productivity was in the long-term, despite increases in tree growth in the early years. Moreover, large N additions resulted in lower soil pH values, with base cations being lost and increasing Al³⁺ concentrations in the soil solution, clearly showing the relationships among biogeochemical processes and the need for long-term studies of them.

An interesting twin approach was used in the early 1980s, when trials started in central Sweden (with Pinus sylvestris; 30 × 30 m²) and Australia (Pinus radiata; 50 × 50 m²), manipulating water and nitrogen supply. Using comparable designs with control, irrigation, and solid or liquid nitrogen addition plots, as well as their respective combination of nitrogen × irrigation plots (Linder 1987), both trials supplied nitrogen related to the daily N demand (in Sweden) or weekly N demand (in Australia) of the trees. Although the trials’ main focus was on the primary production of trees and understorey, soil, tree water use and nutrient dynamics, as well as plant traits (e.g. foliage characteristics) were also measured. Both experiments showed the same general patterns, despite very different environmental settings and actual magnitudes of response rates. The major results included increased tree growth (but only when nitrogen and water were both added to support the larger basal and leaf areas), increased water use efficiency and substantial internal nutrient retranslocation prior to dormancy. Because of the increased tree growth, the canopy microclimate also changed, affecting the understorey and lichen composition, as well as litter decomposition. Thus, emergent properties (Chap. 13, Sect. 13.3) became apparent, resulting in feedbacks and interactions among different traits and processes within these forest stands.

14.2.2.3 Roof Experiments

Another approach to manipulate environmental conditions in ecosystems is the reduction of resource supply—for example, the supply of N or water. By use of large roofs (static or mobile), precipitation and/or atmospheric deposition have been excluded in forest studies since the 1980s (Wright 1989; Gunderson et al. 1998). Questions about the impacts and the reversibility of soil acidification due to S and N deposition could thereby be answered. For example, it was shown that nitrate leaching was reduced rather fast after roof establishment—that is, when N inputs via throughfall were reduced. In addition, nitrogen concentrations in spruce needles decreased in comparison with controls. Although much was learned about the effect of N deposition, questions about nitrogen saturation and critical levels of N deposition for terrestrial ecosystems are still debated today (Binkley and Högberg 2016). Since the 2000s and 2010s, the research focus has shifted and rain-out shelters have been increasingly used to simulate drought conditions in grasslands, arable croplands and forests (Vicca et al. 2014). The results were generally less clear for this resource, since experimental levels established for soil water availability differed more than those, for example, for nitrogen deposition (typically reduced to zero under the roofs). Nevertheless, reduced soil moisture and thus water availability often led to reduced vegetation growth and soil activity, and sometimes also to a change in vegetation composition. Also multifactor climate change manipulations (typically with two- or three-factor combinations of enriched CO₂, increased temperatures and reduced precipitation) have been carried out (Kreyling and Beier 2013; Frank et al. 2015).

14.2.2.4 Free-Air Carbon Dioxide Enrichment Experiments

Triggered by increasing atmospheric CO₂ concentrations measured globally, and interest in their impacts on terrestrial ecosystems, Free-Air Carbon Dioxide Enrichment (FACE) experiments have been carried out since the 1990s in many different ecosystem types, ranging from agroecosystems (arable land and grassland) (Nösberger et al. 2006) to wetlands, deciduous and evergreen forests, and even a desert site (Norby and Zak 2011). However, the overall number of active FACE experiments is currently decreasing because of financial constraints. Some notable exceptions are the recent set-up of EucFACE in Australia (operational since 2013 in a native eucalypt forest) and the plans to set up a FACE experiment in the Amazon (AmazonFACE). These FACE experiments have allowed a new type of ecosystem-scale climate impact study to be performed, enabling scientists to test in situhow ecosystems would respond to elevated atmospheric CO₂ concentrations, going beyond climate chamber, open-top chamber (OTC) and greenhouse studies (Fig. 14.5).

/epubstore/S/E-D-Schulze/Plant-Ecology/OEBPS/images/72100_2_En_14_Chapter/72100_2_En_14_Fig5_HTML.jpg — Fig. 14.5
Different experimental set-ups to simulate enriched atmospheric CO₂ concentrations. a and b Installations in climate chambers and in the greenhouse (Photos courtesy of M. Barthel). c Open-top chambers at the Swiss Federal Institute for Forest, Snow and Landscape Research WSL, Switzerland (Photo courtesy of M. Arend). d and e Free-Air Carbon Dioxide Enrichment (FACE) experiments in Eschikon, Switzerland (Photo courtesy of J. Nösberger) and EucFACE. (Photo courtesy of D. Ellsworth)

The circular plots in most FACE experiments have a diameter of up to 30 m (Hendrey and Miglietta 2006) and are surrounded by pipes that release either pure CO₂ or air enriched with CO₂. The release depends on the wind direction and wind speed, and is thus controlled by a computer system. Flow rates are adjusted to achieve the set target CO₂ concentration (often between 550 and 700 parts per million (ppm)) within the vegetation of each plot, but only during the day (when photosynthesis is taking place) and not at night nor during the leafless period of the year. This set-up allows tall vegetation stands to be studied and avoids many confounding effects of climate chambers, OTCs or greenhouse settings (i.e. size, light intensity, soil conditions). In forest FACE experiments, particularly, mainly young plantations have been studied. At most FACE sites, net primary production (NPP) increased in the first years of operation because of the so-called fertilisation effect, but the growth response to elevated CO₂ concentrations diminished over time, most probably because of physiological down-regulation at the leaf level (Chap. 12), as well as development of nutrient deficiencies (limited soil nitrogen and phosphorus supply), thus interactions at ecosystem level. Therefore, the postulated (short-term) fertilisation effect is not supported by long-term measurements. The Swiss Web-FACE, which released high CO₂ from thin tubes hanging in the tree crowns instead of from tall pipes, overcame the age problem (Körner et al. 2005) but provided information on individual mature deciduous trees (n = 11) rather than on a mature forest stand. After 8 years of operation, this experiment showed no elevated CO₂ effect on the stem growth, litter production or leaf traits of the deciduous trees studied (Bader et al. 2013). However, tree–water relations were affected: the trees transpired less, leading to higher soil moisture levels. Overall, so far, FACE experiments have shown no carbon limitation of the ecosystems under study. Also the new EucFACE supports this conclusion: unexpectedly, the growth and LAI of the eucalypt forest did not respond to elevated CO₂ concentrations in the first 3 years after operation started; it responded only to natural water limitations (Duursma et al. 2016).

The same set-up used for elevated CO₂ research has also been used for ozone (O₃). In Swiss grassland, O₃ concentrations were increased 1.2–1.6 times over ambient levels using circular plots (Volk et al. 2006); in a 60-year-old mixed beech/spruce forest in southern Germany, O₃ concentrations in canopy air were increased by a factor of 2 during the growing season using tubes hanging in the canopy (Matyssek et al. 2010). In both ecosystem types, growth was significantly decreased (by 25% in grassland and by about 44% in mixed forest) because of O₃-induced stomatal closure and thus lower photosynthesis rates in comparison with the ambient controls. In addition, species composition changed in the grassland (with strong reductions in legume abundance) and water run-off increased in the forest, clearly illustrating how important a systems approach is to understand environmental impacts on terrestrial ecosystems.

14.2.2.5 Controls

In each of these manipulation experiments, a new “environment” is created for the ecosystem under study, which is then compared with either ambient conditions or yet another treatment. Preferentially, the plot set-up should thus be similar for all treatments. This means that the control plots in a drought experiment should also have roofs or rain-out shelter structures to keep the canopy microclimate similar; the controls in a soil-warming experiment should also have heating cables installed to create similar soil disturbances for roots and microorganisms in all plots. However, such a set-up will require some additional efforts, since the control plots in a drought experiment need to be supplied with (collected) rainwater in a manner as similar to the natural precipitation patterns as possible. This requires not only a collection facility for precipitation water close by but also pipes and pumps so that water can be supplied to the control plots. The timing and the amount need to be controlled and triggered by the signals of a nearby weather station in near-real time. One can imagine that such additional efforts to create a manipulation infrastructure are often not made; instead, environmental variables in the control and treatment plots are closely followed—for example, by dedicated measurements of the soil water content and VPD.

14.2.2.6 Hidden Treatments

Furthermore, any manipulation of an environmental factor might trigger further changes in the ecosystem, which are most often unintended and sometimes even unknown until much later. A typical example are unintended changes (albeit well known) in light conditions—in terms of both the absolute amount and also the spectral composition—when any kind of roof, foil or shelter material is used. Such changes are due to the material itself and also to shading by the structure carrying the roof or foil or shading by deposits on the roof, foil or shelter surfaces (e.g. dust, pollen). Another example of previously unknown or unexpected interactions comes from a forest FACE experiment. It was recognised only after some years into the experiment that trees growing in the enriched-CO₂ plots depleted the soil nitrogen pool much faster than the control trees and thus experienced slight nitrogen deficiency, counteracting the growth stimulation of higher CO₂ concentrations (Oren et al. 2001). Supplemental N fertilisation triggered the growth of trees growing in the enriched-CO₂ plots to the level of the early years in this FACE experiment. Yet another example from large-roof experiments showed that the sprinkler system could not reproduce the natural variability of rain—in particular, for small rain events—resulting in unintended drying of the litter layer with consequences for organic matter decomposition and mineralisation below the roofs (Gunderson et al. 1998). It is clear that such additional effects cannot be fully avoided, since ecosystems are complex (Mikkelsen et al. 2008). However, an appropriate experimental set-up must be in place to eventually detect these “hidden treatments” and to either experimentally counteract them or take them into account during data analyses and interpretation.

14.2.3 Manipulations of Biodiversity

Studies experimentally manipulating biodiversity have been advocated to overcome the potentially confounding effects described in Sect. 14.1.1. Such experiments allow decoupling of diversity effects from environmental conditions and might enable quantification of causality (for modelling, Chap. 15). Experimental manipulation of diversity also allows isolation of different aspects of biodiversity—for example, the effects of plant species richness, functional diversity or phylogenetic diversity. Two main types of manipulation experiments in biodiversity research have been carried out: removal experiments and synthetic assemblage experiments. Typical research questions asked are: What is the effect of decreased species richness on water/nutrient/carbon dynamics in the ecosystem? Is there a “legacy effect” on plant performance via the soil? Do top-down or bottom-up processes drive biodiversity–ecosystem functioning relationships? Is the diversity of other organisms in the ecosystem linked to the diversity of plants? How are ecosystem processes related to functional diversity?

14.2.3.1 Removal Experiments

In removal experiments, a gradient in diversity levels is created, ranging from natural to depauperate communities, by removing selected components from these ecosystems (e.g. species, functional groups) (Fig. 14.6). These types of experiments mimic the loss of species—for example, as a result of specific pests or pathogens in the past. They reflect natural abiotic and biotic filtering of the regional species pool, represent non-random extinction scenarios (due to the order of species removal as a treatment) and usually involve a large variety of organismic and functional groups (Díaz et al. 2003). However, they face the problem of proper control treatments and often induce large disturbance effects (when above-ground or below-ground tissues are removed), changes in density, spatial segregation of species or increased biogeochemical cycling (when increased root litter is left to decompose). On the other hand, important lessons can be learned. For example, removal of selected plant functional groups and plant species on forested Swedish islands of different sizes (used as a proxy for fire disturbance) affected many ecosystem processes and also clearly demonstrated that the results were ecosystem specific and thus context specific (Wardle and Zackrissen 2005). This means that relationships observed in one ecosystem cannot necessarily be transferred to and used in other ecosystems, unless they are properly tested.

/epubstore/S/E-D-Schulze/Plant-Ecology/OEBPS/images/72100_2_En_14_Chapter/72100_2_En_14_Fig6_HTML.png — Fig. 14.6
Comparison of complementary scientific approaches to the study of biodiversity–ecosystem functioning relationships (based on Díaz (2013)). Modelling studies are not included. The examples shown in the photographs are from a comparative study in subtropical China (*left*) (Bruelheide et al. 2011; Photo: M. Scherer-Lorenzen), a species removal experiment (Photo courtesy of C. Dorman) and the Jena Experiment in Germany (*right*). (Roscher et al. 2004; Photo courtesy of the Jena Experiment)

14.2.3.2 Synthetic Assemblage Experiments

Establishing new communities, according to self-set rules, such synthetic assemblage experiments have been increasingly used over the last two decades to study biodiversity–ecosystem functioning relationships, particularly for plant and microbial diversity studies. For plant diversity studies, a biodiversity gradient is created by sowing or planting, keeping environmental conditions (e.g. climate, fertility and land use history) as constant as possible (Fig. 14.6). Such experiments are conducted in the field and also in controlled environmental facilities (Sect. 14.2.5). For very practical reasons, fast-growing, small-sized, mainly early-successional model systems are used, often with grassland species. Nevertheless, also tree diversity is being studied. Probably the largest research facility for ecosystem science is the global network for tree diversity experiments, TreeDivNet (www.treedivnet.ugent.be), with one million trees planted for science at 36 sites totalling 800 ha and 4000 plots (Verheyen et al. 2016). The first experiments in the Ecotron facility at Imperial College/Silwood Park, UK (a multitrophic study) (Naeem et al. 1994) in North American prairie systems at Cedar Creek, MN, USA (Tilman et al. 1996) and in serpentine grasslands in California, USA (Hooper and Vitousek 1997), as well as in European grasslands in the BIODEPTH project (Hector et al. 1999) paved the road for even larger and more sophisticated designs. The second generation of biodiversity experiments included larger plot sizes (up to 20 × 20 m in grassland) (Roscher et al. 2004) and greater replication of diversity treatments (e.g. Tilman et al. 1997). They were designed to allow separation of sampling from complementarity effects (Chap. 20) and study of multitrophic interactions, as well as interactions with land use intensity and drought (e.g. the Jena Experiment) (Roscher et al. 2004; Weigelt et al. 2009; Vogel et al. 2012). The most prominent result was probably the positive relationship between species richness and productivity, which was highly consistent across all experiments (for details, Chap. 20). Some diversity experiments also allow the study of interactions with other global change drivers, such as nitrogen deposition and increasing CO₂ levels (e.g. Reich et al. 2001). The drawbacks of these synthetic assemblage experiments have been discussed intensively—for example, artefacts introduced by certain experimental procedures (e.g. the need to continuously weed unsown species), often unrealistic random draw diversity loss scenarios or the occurrence of transient effects (Díaz et al. 2003; Lepš 2004). Nevertheless, these experiments resulted in a profound knowledge gain about the way ecosystems work (Chap. 20).

14.2.4 Manipulations of Management and Changes in Land Cover

Terrestrial ecosystems are often managed ecosystems, as introduced in Chap. 13. Thus, it does not come as a surprise that some of the oldest whole-ecosystem experiments focus on agricultural systems and study the relationships of the management regime (e.g. different fertilisation schemes) and land use change (e.g. comparison of mature forest stands and clear-cuts).

The Park Grass Experiment (about 2.8 ha in area) (Fig. 14.7) at Rothamsted Experimental Station (now Rothamsted Research), Harpenden, UK, started in 1856 to study how different fertilisers affect the biomass yield from grasslands. It is the oldest permanent grassland experiment worldwide (https://www.rothamsted.ac.uk/long-term-experiments). After previous use of the site as a pasture for at least a century, different treatments with inorganic fertilisers were established at the very start (nitrate-N, ammonium-N, P, K, Mg, Na and Si) while treatments with organic fertilisers were added 50 years later (in 1905, farmyard manure (FYM) and fishmeal). Since the start of the experiment in the nineteenth century, the plots have been decreased in size twice (in 1903 and 1965) to accommodate different liming treatments (control and set pH values of 5, 6 and 7). The extensive agricultural management has stayed the same over the last 160 years, with plots being mown and cut plant tissues being removed in summer (typically in June) and autumn.

/epubstore/S/E-D-Schulze/Plant-Ecology/OEBPS/images/72100_2_En_14_Chapter/72100_2_En_14_Fig7_HTML.png — Fig. 14.7
The Park Grass Experiment, the oldest permanent grassland experiment on Earth, established in 1856. a Aerial view. b Plot layout. (Photo courtesy of Rothamsted Research)

Early on, it became clear that management practices and soil pH were closely related to biomass production and also to plant species richness. Over time, evidence arose that community composition and also soil chemistry and biology responded negatively to atmospheric N deposition. The experiment provided clear evidence for small-scale adaptation (at the plot level) and reproductive isolation among plots by natural selection. Higher trophic levels were also studied and showed treatment-specific distribution among the different plots (for a review of results for 1856–2006, see Silvertown et al. 2006). Thus, this grassland experiment clearly illustrates the value long-term experiments have, and also their ever increasing value the longer they run. It has also become clear over the last 160 years that experiments set up originally for a very different purpose (here, to study the response to management) might be used over time to answer completely different research questions from those originally posed, particularly when new technology becomes available (Chap. 13).

Manipulations in management are typically not followed over such long periods. However, the resampling of older experiments can also yield highly valuable results. One fine example is a study by Spiegelberger et al. (2006). In 2002, they resampled a large-scale fertilisation experiment in a subalpine grassland near Interlaken in Switzerland, which had been set up in the 1930s (1932–1935) as a multifactorial experiment with 340 plots to study the effects of liming and NPK fertilisation. Although these treatments had been applied for only 2–4 years, the plant community composition and also the soil microbial community showed clear effects of the liming treatments, though not the NPK treatments, after almost 70 years! The authors concluded that even such short-term changes in management can have profound and long-lasting impacts—here, on soil pH—indicating very low resilience of these mountain grasslands.

The oldest experiment manipulating entire forest ecosystems is the Hubbard Brook Ecosystem Study (HBES) (Fig. 14.8) (Holmes and Likens 2016). Established in 1955 to investigate management of watersheds in New England, experiments in the Hubbard Brook Experimental Forest (HBEF) started in 1963. Neighbouring small watersheds with northern hardwood forest were assigned to different treatments: control (watershed 6), clear-cut in winter 1965 followed by herbicide treatment (1965–1968; watershed 2) and further logging experiments (watershed 5; whole-tree harvest). Use of a “small watershed technique” (Likens et al. 1977) allowed measurements of both the volume and the chemistry of precipitation inputs and stream water outputs of a spatially clearly defined unit—that is, a small watershed (Chap. 13)—and to study the effects of changing land cover on ecosystem nutrient budgets. Since then, many more experiments and observations have been carried out within the watershed of the Hubbard Brook (www.hubbardbrook.org). Of the original experiments, in particular, the clear-cut of an entire forested watershed and the subsequent herbicide treatment to prevent regrowth for 3 years provided unprecedented insight into the responses of watershed hydrology and biogeochemistry to land cover changes. Streamflow increased by a factor of 5 in the following year (hydrological year 1965–1966), and evapotranspiration decreased by a factor of 4. Even more spectacular, nitrate concentrations in stream water increased from close to zero to values of about 50 mg/L and stayed high, similar to cation concentrations, until regrowth of vegetation started in 1969. These changes in water and nutrient pools and fluxes, related to vegetation cover, were later shown to also affect other ecosystem characteristics and processes, such as surface albedo, C sink strengths, erosion and biological diversity, to name a few (see Holmes and Likens (2016) for the latest synthesis).

/epubstore/S/E-D-Schulze/Plant-Ecology/OEBPS/images/72100_2_En_14_Chapter/72100_2_En_14_Fig8_HTML.jpg — Fig. 14.8
Hubbard Brook Experimental Forest. a Map of the watershed (Holmes and Likens 2016; copyright granted by Yale University Press). b Map of experimental watersheds (watershed 2: clear-cut; watershed 5: whole-tree harvest; watershed 6: control). (Photo courtesy of Hubbard Brook Experimental Forest)

14.2.5 Artificial Ecosystems

A very different approach to the study of entire ecosystems from the ones mentioned above is to “create” entire ecosystems artificially under controlled or semi-controlled environments. Different facilities of various sizes are available (micro-, meso- and macrocosms) and have been used in recent decades to answer research questions such as: How do changes in environmental conditions affect the root exudation in diverse grasslands? How do nutrient and water fluxes interact and affect the NPP of different community members in a given ecosystem?

Open-top chambers (OTCs) and whole-tree chambers (WTCs): These mesocosms are typically round enclosures of varying heights (from <1 m to about 10 m) and diameters (from <1 m to about 3 m), depending on the ecosystem type under study (Fig. 14.9; OTC). They are covered by a transparent dome made of plastic, foils or films with high transmission of UV and visible light. Often they are open to the top/side to allow natural precipitation to enter and, to prevent heating up, sometimes they are temperature controlled. Typically, OTCs have a confined below-ground compartment as well, either with walls extending into the native soil or with more sophisticated instrumentation such as weighing devices or lysimeters filled with soil (see below). Depending on the set-up, OTCs contain either local vegetation on grown soil or planted/seeded communities in lysimeters filled with soil. For example, OTCs treated with single and combinations of global change drivers (elevated CO₂, increased soil and air temperatures, increased N deposition, increased precipitation) were used in the 1990s in the Jasper Ridge Global Change Experiment (http://globalecology.stanford.edu/DGE/Dukes/JRGCE/home.html) to study the responses of Californian grassland to global environmental change (Shaw et al. 2002).
Lysimeters: These are cylinders, open at the top and closed at the bottom, with diameters and heights of up to about 3 m. Typically put into the ground so they form an even surface with the surrounding vegetation, they are filled with (in the best case) undisturbed soil and covered by sown/planted vegetation. To study soil water dynamics and soil leaching, the drainage water in the cylinder is collected at the bottom and used for further chemical analyses. Lysimeters can be combined with other facilities, such as OTCs or WTCs, or put onto balances. Then they require underground walk-in facilities for maintenance and sampling. One prime example for such a combined set-up is the Montpellier European Ecotron in Montferrier-sur-Lez, France (www.ecotron.cnrs.fr). Here, 30 m³ transparent domes are situated directly on top of lysimeters that are 1.6 m in diameter and 2 m in depth (Fig. 14.9).

/epubstore/S/E-D-Schulze/Plant-Ecology/OEBPS/images/72100_2_En_14_Chapter/72100_2_En_14_Fig9_HTML.png — Fig. 14.9
Macrocosm facility at the Montpellier European Ecotron. a Scheme of the set-up: an intact ecosystem is placed below a dome made of Teflon-FEP (fluorinated ethylene propylene) film. The soil monolith is placed in a lysimeter, which is located in an underground, walk-in, air-conditioned room. The air is circulated through the soil and air-diffusing rings into the dome, and it leaves the dome through circular pipes at the soil surface. b Setting up a new experiment with soil monoliths. c View of the lysimeter room, with the weighing system and many soil sensors for measurements in the soil monolith. d View of a sown grassland situated in the lysimeter. The size of the ecosystem can be adjusted in diameter and soil depth, here seen by the green metal ring. (Photos courtesy of Montpellier European Ecotron)

Biosphere 2: A well-known example of an artificial ecosystem is Biosphere 2 (named after “Biosphere 1”—the Earth), which was established close to Tucson, AZ, USA, in 1991 to test whether human life would be possible in a closed, self-sufficient ecological system; if successful it was to be used as a role model for longer space missions. The facility (1.3 ha in area and about 204000 m³ in volume) included about 3800 species in several ecosystems (ocean, mangrove, tropical rainforest, savanna, desert, intensive agriculture) and living quarters for eight persons, who stayed in Biosphere 2 for about 2 years. However, the first experiment (1991–1993) failed because of decreases (to about 14%) in the concentrations of oxygen (which was slowly consumed by microbial respiration in soils, while respiratory CO₂ was captured by the concrete structures) (Severinghaus et al. 1994) and large changes in species abundance (e.g. loss of pollinators versus dominance of cockroaches and greenhouse ants), nicely demonstrating the complexity of ecological interactions. The second experiment (in 1994) was terminated after 6 months because of a management controversy. The facility has subsequently been used for ecological and biogeochemical research by the University of Columbia (1996–2003) and the University of Arizona, which has owned the Biosphere 2 Laboratory (B2L) since 2011. The facility has been used to carry out global change experiments—for example, with increasing CO₂ concentrations, air and soil temperature manipulations and drought simulations—as well as to test new technological approaches (Gonzalez-Meler et al. 2014). B2L closes the methodological gap between small enclosures (such as OTCs, WTCs and small growth chambers) and mature ecosystems (such as forest stands) and, for example, has enabled the study of isoprene emissions in ecosystems under controlled conditions and provided data to improve the representation of soil respiration in ecosystem models. Today, B2L is still used for research, and also for education and outreach (http://biosphere2.org/). Although not the first artificial “Biosphere” (Box 14.1), B2L is clearly the most famous one.

Box 14.1: Bios-1 to Bios-3

Bios-1 to Bios-3 (sometimes also called CELSS (for Controlled/Closed Ecological/Environment Life-Support System)), designed for the Soviet space programme and employed until 1984 in experiments involving humans, had a mission very similar to that of Biosphere 2. Using highly artificial and strongly simplified ecosystems, Biosphere 3 was employed to find a way to sustain human life in outer space for a long period of time. Although now named Biosphere 3, it was designed and used by the Institute for Biophysics in Krasnojarsk, Siberia, much earlier (1965 and 1972) than Biosphere 2 but became known only much later because of the Cold War. Biosphere 3 was smaller than Biosphere 2 (only 315 m³ in volume) and was made to accommodate a maximum of three persons at a time. The longest experiment ran for 180 days. Waste recycling using green algae (Chlorella vulgaris) or higher green plants (wheat, sedge nuts and vegetables) allowed removal of carbon dioxide in the system while producing oxygen and food. Three experiments, always with three persons in the facility, were carried out with different set-ups (algae versus higher plants, differences in duration, differences in crew composition), all providing great insights into the theory of closed systems and how to model them. Important aspects discovered during these experiments were a certain lack of self-regulation, the challenge of balancing food production and food consumption, the role of trophic levels in human diets, the need to avoid “deadlock substances” (i.e. elements and molecules, particularly micronutrients, being permanently removed from the system) and changes in microbial populations both in the soils and on human skin (Salisbury et al. 1997). Ensuring “stability” in any of these artificial systems was a difficult task in both these small systems and the larger Biosphere 2, demonstrating that stability or resilience can maybe only be achieved in rather complex, diverse ecosystems.

14.3 Summary

Many different approaches to study terrestrial ecosystems are available.
Observations include whole-ecosystem studies, transects, chronosequences, grid-based inventories, remote sensing applications and natural experiments.
Experiments include manipulations of pools and fluxes, of environmental conditions and biodiversity, and also of management and land cover. Artificial ecosystems can be designed in open-top chambers, whole-tree chambers or macrocosms such as Biosphere 2. Selection of appropriate controls is essential—for example, to account for hidden treatment effects.
Whole-ecosystem studies are often part of larger networks, increasing spatial representativeness and also enabling standardisation of measurements and data processing.
Development of new technologies—for example, for canopy access or measurements of trace gas fluxes—has increased our understanding of ecosystems dramatically.
Long-term studies avoid the problem of transient effects. They become more valuable the longer they run.

References

Alexander JM, Diez JM, Levin JM (2015) Novel competitors shape species’ responses to climate change. Nature 525:515–520PubMed
Allen CD, Macalady AK, Chenchouni H, Bachelet D, McDowell N, Vennetier M, Kitzberger T, Rigling A, Breshears DD, Hogg EH, Gonzalez P, Fensham R, Zhangm Z, Castro J, Demidova N, Lim JH, Allard G, Running SW, Semerci A, Cobb N (2010) A global overview of drought and heat-induced tree mortality reveals emerging climate change risks for forests. For Ecol Manag 259:660–684
Allen CD, Breshears DD, McDowell NG (2015) On underestimation of global vulnerability to tree mortality and forest die-off from hotter drought in the Anthropocene. Ecosphere 6:129
Allison SD, Vitousek PM (2004) Rapid nutrient cycling in leaf litter from invasive plants in Hawai’i. Oecologia 141:612–619PubMed
Anderson K, Gaston KJ (2013) Lightweight unmanned aerial vehicles will revolutionise spatial ecology. Front Ecol 11:138–146
Assmann E, Franz F (1963) Voriaufige Fichtenertragstafeln fur Bayem. Mittleres Ertragsniveau. In: Bayerisches Staatsministerium fiir Emahrung, Landwirtschaft und Forsten (eds) Hilfstafeln fur die Forsteinrichtung. Bayerisches Staatsministerium fur Emahrung, Landwirtschaft und Forsten, Munich
Aubinet M, Vesala T, Papale D (eds) (2012) Eddy covariance: a practical guide to measurement and data analysis. Springer Atmospheric Sciences, Springer Verlag, Dordrecht
Bader MKF, Leuzinger S, Keel SG, Siegwolf RTW, Hagedorn F, Schleppi P, Körner C (2013) Central European hardwood trees in a high-CO₂ future: synthesis of an 8-year forest canopy CO₂ enrichment project. J Ecol 101:1509–1519
Baeten L, Verheyen K, Wirth C, Bruelheide H, Bussotti F, Finér L, Jaroszewicz B, Selvi F, Valladares F, Allan E, Ampoorter E, Auge H, Avăcăriței D, Barbaro L, Bărnoaiea I, Bauhus J, Beinhoff C, Benavides R, Benneter A, Berger S, Berthold F, Boberg J, Bonal D, Brüggemann W, Carnol M, Castagneyrol B, Charbonnier Y, Chećko E, Coomes D, Coppi A, Crespo BC, Dalmaris E, Dănilă G, Dawud SM, de Vries W, De Wandeler H, Deconchat M, Domisch T, Duduman G, Fischer M, Fotelli M, Gessler A, Gimeno TE, Granier A, Grossiord C, Guyot V, Hantsch L, Hättenschwiler S, Hector A, Hermy M, Holland V, Jactel H, Joly FX, Jucker T, Kolb S, Koricheva J, Lexer MJ, Liebergesell M, Milligan H, Müller S, Muys B, Nguyen D, Nichiforel L, Pollastrini M, Proulx R, Rabasa S, Radoglou K, Ratcliffe S, Raulund-Rasmussen K, Seiferling I, Stenlid J, Vesterdal L, von Wilpert K, Zavala MA, Zielinski D, Scherer-Lorenzen M (2013) A novel comparative research platform designed to determine the functional significance of tree species diversity in European forests. Perspect Plant Ecol Evol Syst 15:281–291
Bardgett RD, Wardle DA (2010) Aboveground–below-ground linkages, Series in ecology and evolution. Oxford University Press, Oxford
Barker R (2005) Scorched earth: how the fires of Yellowstone changed America. Island Press, Washington, DC
Bassin S, Volk M, Suter M, Buchmann N, Fuhrer J (2007) Nitrogen deposition but not ozone affects productivity and community composition of sub-alpine grassland after three years of treatment. New Phytol 175:523–534PubMed
Beier C, Beierkuhnlein C, Wohlgemuth T, Penuelas J, Emmett B, Körner C, de Boeck H, Hesselbjerg Christensen J, Leuzinger S, Janssens IA, Hansen K (2012) Precipitation manipulation experiments—challenges and recommendations for the future. Ecol Lett 15:899–911PubMed
Binkley D, Högberg P (2016) Tamm review: revisiting the influence of nitrogen deposition on Swedish forests. For Ecol Manag 368:222–239
Bruelheide H, Böhnke M, Both S, Fang T, Assmann T, Baruffol M, Bauhus J, Buscot F, Chen X, Ding BY, Durka W, Erfmeier A, Fischer M, Geißler C, Guo D, Guo LD, Härdtle W, He JS, Hector A, Kröber W, Kühn P, Lang AC, Nadrowski K, Pei KQ, Scherer-Lorenzen M, Shi XZ, Scholten T, Schuldt A, Trogisch S, von Oheimb G, Welk E, Wirth C, Wu YT, Yang XF, Yu MJ, Zeng XQ, Zhang SR, Zhou HZ, Ma KP, Schmid B (2011) Community assembly during secondary forest succession in a Chinese subtropical forest. Ecol Monogr 81:25–41
Brumme R, Khanna PK (2009) Functioning and management of European beech ecosystems, Ecological Studies, Vol 208. Springer, Berlin
Buchmann N (2002) Plant ecophysiology and forest response to global change. Tree Physiol 22:1177–1184PubMed
Buchmann N, Schulze ED (1999) Net CO₂ and H₂O fluxes of terrestrial ecosystems. Glob Biogeochem Cycles 13:751–760
Burri S, Sturm P, Prechsl UE, Knohl A, Buchmann N (2014) The impact of extreme summer drought on the short-term carbon coupling of photosynthesis to soil CO₂ efflux in a temperate grassland. Biogeosciences 11:961–975
Canadell JG, Steffen WL, White PS (2002) IGBP/GCTE terrestrial transects: dynamics of terrestrial ecosystems under environmental change. J Veg Sci 13:297–450
Coleman DC (2010) Big ecology: the emergence of ecosystem science. University of California Press, Berkeley
Diamond JM (1983) Laboratory, field and natural experiments. Nature 304:586–587
Díaz S (2013) Ecosystem function measurement, terrestrial communities. In: Simon AL (ed) Encyclopedia of biodiversity, 2nd edn. Academic Press, Waltham, pp 72–89.
Díaz S, Symstad AJ, Stuart Chapin IF, Wardle DA, Huenneke LF (2003) Functional diversity revealed by removal experiments. Trends Ecol Evol 18:140–146
Duursma RA, Gimeno TE, Boer MM, Crous KY, Tjoelker MG, Ellsworth DS (2016) Canopy leaf area of a mature evergreen Eucalyptus woodland does not respond to elevated atmospheric [CO₂] but tracks water availability. Glob Change Biol 22:1666–1676PubMed
Ellenberg J (1971) Integrated experimental ecology, Ecological Studies, Vol 2. Springer, Berlin
Frank D, Reichstein M, Bahn M, Thonicke K, Frank D, Mahecha M, Smith P, van der Velde M, Vicca S, Babst F, Beer C, Buchmann N, Canadell JG, Ciais P, Cramer W, Ibrom A, Miglietta F, Poulter B, Rammig A, Seneviratne SI, Walz A, Wattenbach M, Zavala MA, Zeischler J (2015) Effects of climate extremes on the terrestrial carbon cycle: concepts, processes and potential future impacts. Glob Change Biol 21:2861–2880PubMedPubMedCentral
Franke MA (2000) Yellowstone in the afterglow: lessons from the fires. National Park Service, Mammoth Hot Springs
French NR (1979) Perspectives in grassland ecology: results and applications of the US/IBP grassland biome study. Ecological Studies: analysis and synthesis, Vol 32. Springer, New York
Gonzalez-Meler MA, Rucks JS, Aubanell G (2014) Mechanistic insights on the responses of plant and ecosystem gas exchange to global environmental change: lessons from Biosphere 2. Plant Sci 226:14–21PubMed
Greenaway F (1996) Science international: a history of the International Council of Scientific Unions. Cambridge University Press, Cambridge
Guanter L, Zhang Y, Jung M, Joiner J, Voigt M, Berry JA, Frankenberg C, Huete A, Zarco-Tejada P, Lee J-E, Moran S, Ponce-Campos G, Beer C, Camps-Valls G, Buchmann N, Gianelle D, Klumpp K, Cescatti A, Baker JM, Griffis TJ (2014) Global and time-resolved monitoring of crop photosynthesis with chlorophyll fluorescence. Proceedings of the National Academy of Sciences 111:E1327–E1333
Gunderson P, Boxman AW, Lamersdrof N, Moldan F, Andersen BR (1998) Experimental manipulation of forest ecosystems: lessons from large roof experiments. For Ecol Manag 101:339–352
Hallé F, Cleyet-Marrel D, Ebersolt G (2000) Le Radeau des Cimes: l’exploration des canopées forestières. JC Lattès, Paris
Hector A, Schmid B, Beierkuhnlein C, Caldeira MC, Diemer M, Dimitrakopoulos PG, Finn J, Freitas H, Giller PS, Good J, Harris R, Högberg P, Huss-Danell K, Joshi J, Jumpponen A, Körner C, Leadley PW, Loreau M, Minns A, Mulder CPH, O’Donovan G, Otway SJ, Pereira JS, Prinz A, Read DJ, Scherer-Lorenzen M, Schulze ED, Siamantziouras ASD, Spehn E, Terry AC, Troumbis AY, Woodward FI, Yachi S, Lawton JH (1999) Plant diversity and productivity experiments in European grasslands. Science 286:1123–1127PubMed
Hendrey GR, Miglietta F (2006) FACE technology: past, present, and future. Ecol Stud 187:15–45
Hiller RV, Bretscher D, DelSontro T, Diem T, Eugster W, Henneberger R, Hobi S, Hodson E, Imer D, Kreuzer M, Künzle T, Merbold L, Niklaus PA, Rihm B, Schellenberger A, Schroth MH, Schubert CJ, Siegrist H, Stieger J, Buchmann N, Brunner D (2014) Anthropogenic and natural methane fluxes in Switzerland synthesised within a spatially-explicit inventory. Biogeosciences 11:1941–1960
HiIleRisLambers J, Ettinger AK, Ford KR, Haak DC, Horwith M, Miner BE, Rogers HS, Sheldon KS, Tewksbury JJ, Waters SM, Yang S (2013) Accidental experiments: ecological and evolutionary insights and opportunities derived from global change. Oikos 122:1649–1661
Högberg P (2001) Interactions between hillslope hydrochemistry, nitrogen dynamics and plants in Fennoscandian boreal forest. In: Schulze E-D, Heimann M, Harrison S, Holland E, Lloyd J, Prenice IC, Schimel D (eds) Global biogeochemical cycles in the climate system. Academic Press, San Diego, pp 227–233
Högberg P, Nordgren A, Buchmann N, Taylor AFS, Ekblad A, Högberg MN, Nyberg G, Ottosson-Löfvenius M, Read DJ (2001) Large-scale forest girdling shows that current photosynthesis drives soil respiration. Nature 411:789–792PubMed
Högberg P, Fan H, Quis M, Binkley D, Tamm CO (2006) Tree growth and soil acidification in response to 30 years of experimental nitrogen loading on boreal forest. Glob Change Biol 12:489–499
Högberg P, Högberg MN, Göttlicher SG, Betson NR, Keel SG, Metcalfe DB, Campbell C, Schindlbacher A, Hurry V, Lundmark T, Linder S, Näsholm T (2008) High temporal resolution tracing of photosynthate carbon from the tree canopy to forest soil microorganisms. New Phytol 177:220–228PubMed
Holmes R, Likens GE (2016) Hubbard Brook: the story of a forest ecosystem. Yale University Press, New Haven
Hooper DU, Vitousek PM (1997) The effects of plant composition and diversity on ecosystem processes. Science 277:1302–1305
Ineson P, Taylor K, Harrison AF, Poskitt J, Benham DG, Tipping E, Woof C (1998) Effects of climate change on nitrogen dynamics in upland soils. 1. A transplant approach. Glob Change Biol 4:143–152
Jenny H (1941) Factors of soil formation. A system of quantitative pedology. McGraw-Hill, New York
Kira T, Shidei T (1967) Primary production and turnover of organic matter in different forest ecosystems of the western Pacific. Jpn J Ecol 17:70–87
Kolari P, Pumpanen J, Rannik U, Ilvesniemi H, Hari P, Berninger F (2004) Carbon balance of different aged scots pine forests in southern Finland. Glob Change Biol 10:1106–1119
Körner C, Asshoff R, Bignucolo O, Hättenschwiler S, Keel SG, Pelaez-Riedl S, Pepin S, Siegwolf RTW, Zotz G (2005) Carbon flux and growth in mature deciduous forest trees exposed to elevated CO₂. Science 309:1360–1362PubMed
Kreyling J, Beier C (2013) Complexity in climate change manipulation experiments. Bioscience 63:763–767
Kuzyakov Y, Gavrichkova O (2006) Time lag between photosynthesis and carbon dioxide efflux from soil: a review of mechanisms and controls. Glob Change Biol 16:3386–3406
Lepš J (2004) What do the biodiversity experiments tell us about consequences of plant species loss in the real world? Basic Appl Ecol 5:529–534
Leuschner C, Jungkunst HF, Fleck S (2009) Functional role of forest diversity: pros and cons of synthetic stands and across-site comparisons in established forests. Basic Appl Ecol 10:1–9
Likens GE, Bormann FH, Pierce RS, Eaton JS, Johnson NM (1977) Biogeochemistry of a forested ecosystem. Springer, New York
Linder S (1987) Responses to water and nutrients in coniferous ecosystems. Ecol Stud 61:180–200
Lowman MD (1999) Life in the treetops—adventures of a woman in field biology. Yale University Press, New Haven
Magnani F, Mencuccini M, Borghetti M, Berbigier P, Berninger F, Delzon S, Grelle A, Hari P, Jarvis PG, Kolari P, Kowalski AS, Lankreijer H, Law BE, Lindroth A, Loustau D, Manca G, Moncrieff JB, Rayment M, Tedeschi V, Valentini R, Grace J (2007) The human footprint in the carbon cycle of temperate and boreal forests. Nature 447:848–852
Matyssek R, Wieser G, Ceulemans R, Rennenberg H, Pretzsch K, Häberle KH, Löw M, Nunn AJ, Werner H, Wipfler P, Oßwald W, Nikolova P, Hanke DE, Kraigher H, Tausz M, Bahnweg G, Kitao M, Dieler J, Sandermann H, Herbinger K, Grebenc T, Blumenröther M, Deckmyn G, Grams TEE, Heerdt C, Leuchner M, Fabian P, Häberle KH (2010) Enhanced ozone strongly reduces carbon sink strength of adult beech (Fagus sylvatica)—resume from the free-air fumigation study at Kranzberg Forest. Environ Pollut 158:2527–2532PubMed
Mikkelsen TN, Beier C, Jonasson S, Holmstrup M, Schmidt IK, Ambus P, Pilegaard K, Michelsen A, Albert K, Andresen LC, Arndal MF, Bruun N, Christensen S, Danbæk S, Gundersen P, Jørgensen P, Kongstad J, Maraldo K, Priemé A, Riis-Nielsen T, Ro-Poulsen H, Stevnbak K, Selsted MB, Sørensen P, Larsen KS, Carter MS, Martinussen T, Miglietta F, Sverdrup H (2008) Experimental design of multifactor climate change experiments with elevated CO₂, warming and drought—the CLIMAITE project. Funct Ecol 22:185–195
Mund M, Kummetz E, Hein M, Bauer GA, Schulze E-D (2002) Growth and carbon stocks of a spruce forest chronosequence in central Europe. For Ecol Manag 171:275–296
Myneni RB, Keeling CD, Tucker CJ, Asrar G, Nemani RR (1997) Increased plant growth in the northern high latitudes from 1981 to 1991. Nature 386:698–702
Naeem S, Bunker DE, Hector A, Loreau M, Perrings C (eds) (2009) Biodiversity, ecosystem functioning, & human wellbeing: an ecological and economic perspective. Oxford University Press, Oxford
Naeem S, Thompson LJ, Lawler SP, Lawton JH, Woodfin RM (1994) Declining biodiversity can alter the performance of ecosystems. Nature 368:734–737
Norby RJ, Zak DR (2011) Ecological lessons from free-air CO₂ enrichment FACE experiments. Annu Rev Ecol Evol Syst 42:181–203
Nösberger J, Long SP, Norby RJ, Stitt M, Hendrey GM, Blum H (2006) Managed ecosystems and CO₂: case studies, processes, and perspectives, Ecological Studies, Vol 187. Springer, Berlin
Odum E (1971) Fundamentals of ecology. WB Saunders, Co., Philadelphia
Odum E (1983) Basic ecology. Harcourt Brace College Publishers, New York
Odum E (2005) Fundamentals of ecology. Thomson Brooks/Cole, Belmont
Oren R, Ellsworth DS, Johnsen KH, Phillips N, Ewers BE, Maier C, Schäfer KVR, McCarthy H, Hendrey G, McNulty SG, Katul GG (2001) Soil fertility limits carbon sequestration by forest ecosystems in a CO₂-enriched atmosphere. Nature 411:469–472PubMed
Ramanathan V (1988) The greenhouse theory of climate change: a test by an inadvertent global experiment. Science 240:293–299PubMed
Reich PB, Knops J, Tilman D, Craine J, Ellsworth D, Tjoekler M, Lee T, Wedin D, Naeem S, Bahauddin D, Hendrey G, Jose S, Wrage K, Goth J, Bengtson W (2001) Plant diversity enhances ecosystem responses to elevated CO₂ and nitrogen deposition. Nature 410:809–812PubMed
Roscher C, Schumacher J, Baade J, Wilcke W, Gleixner G, Weisser WW, Schmid B, Schulze ED (2004) The role of biodiversity for element cycling and trophic interactions: an experimental approach in a grassland community. Basic Appl Ecol 5:107–121
Ruehr NK, Offermann CA, Gessler A, Winkler BJ, Ferrio JP, Buchmann N, Barnard RL (2009) Drought effects on allocation of recent carbon: from beech leaves to soil CO₂ efflux. New Phytol 184:950–961
Running SW, Nemani RR Heinsch FA, Zhao M, Reeves M, Hashimoto H (2004) A continuous satellite-derived measure of global terrestrial primary production. Bioscience 54:547–560
Salisbury FB, Gitelson JI, Lisovsky GM (1997) Bios-3: Siberian experiments in bioregenerative life support. BioScience 47:575–585PubMed
Schulze ED, Mooney HA, Sala OE, Jobbagy E, Buchmann N, Bauer G, Canadell J, Jackson RB, Loreti J, Oesterheld M, Ehleringer JR (1996) Rooting depth, water availability, and vegetation cover along an aridity gradient in Patagonia. Oecologia 108:503–551
Severinghaus JP, Broecker W, Dempster W, MacCallum T, Wahlen M (1994) Oxygen loss in Biosphere 2. Eos Trans AGU 75:35–37
Shaw MB, Zavaleta ES, Chiariello NR, Cleland EE, Mooney HA, Field CB (2002) Grassland responses to global environmental changes suppressed by elevated CO₂. Science 298:1987–1990PubMed
Shugart HH, Macko SA, Lesolle P, Szuba TA, Mukelabai MM, Dowty P, Swap RJ (2004) The SAFARI 2000—Kalahari transect wet season campaign of year 2000. Glob Change Biol 10:273–280
Silvertown J, Poulton P, Johnston E, Edwards G, Heard M, Biss PM (2006) The Park Grass Experiment 1856–2006: its contribution to ecology. J Ecol 94:801–814
Spiegelberger T, Hegg O, Matthies D, Hedlund K, Schaffner U (2006) Long-term effects of short-term perturbation in a subalpine grassland. Ecology 87:1939–1944PubMed
Stork NE, Wright SJ, Mulkey SS (1997) Craning for a better view: the Canopy Crane Network. Trends Ecol Evolut 12:415–420
Sutton SL (2001) Alice grows up: canopy science in transition from Wonderland to reality. Plant Ecol 153:13–21
Tilman D, Wedin D, Knops J (1996) Productivity and sustainability influenced by biodiversity in grassland ecosystems. Nature 379:718–720
Tilman D, Knops J, Wedin D, Reich P, Ritchie M, Siemann E (1997) The influence of functional diversity and composition on ecosystem processes. Science 277:1300–1302
Ulrich B (1983) Interaction of forest canopies with atmospheric constituents: SO₂, alkali and earth alkali cations and chloride. In: Ulrich B, Pankrath J (eds) Effects of accumulation of air pollutants in forest ecosystems. D Reidel Publishing Company, Dordrecht, pp 33–45
Verheyen K, Vanhellemont M, Auge H, Baeten L, Baraloto C, Barsoum N, Bilodeau-Gauthier S, Bruelheide H, Castagneyrol B, Godbold D, Haase J, Hector A, Jactel H, Koricheva J, Loreau M, Mereu S, Messier C, Muys B, Nolet P, Paquette A, Parker J, Perring M, Ponette Q, Potvin C, Reich P, Smith A, Weih M, Scherer-Lorenzen M (2016) Contributions of a global network of tree diversity experiments to sustainable forest plantations. Ambio 45:29–41PubMed
Vicca S, Bahn M, Estiarte M, Alberti G, Ambus P, Arain A, Beier C, Bentley LP, Borken W, Buchmann N, Collins SL, Costa ACL, de Dato G, Dukes JS, Escolar C, Fay P, Guidolotti G, Hanson PJ, Kahmen A, Kröel-Dulay G, Ladreiter-Knauss T, Larsen KS, Lellei-Kovacs E, Lebrija-Trejos E, Maestre FT, Marhan S, Marshall M, Meir P, Miao Y, Muhr J, Niklaus PA, Ogaya R, Peñuelas J, Poll C, Rustad LE, Savage K, Schindlbacher A, Schmidt IK, Smith AR, Sotta ED, Suseela V, Tietema A, van Gestel N, van Loon EE, van Straaten O, Vargas R, Wan S, Weber U, Janssens IA (2014) Can current moisture responses predict soil respiration under altered precipitation regimes? A synthesis of manipulation experiments. Biogeosciences 11:2991–3013
Vilà M, Vayreda J, Comas L, Ibáñez J, Mata T, Obón B (2007) Species richness and wood production: a positive association in Mediterranean forests. Ecol Lett 10:241–250PubMed
Vitousek PM, Aplet G, Turner D, Lockwood JJ (1992) The Mauna Loa environmental matrix: foliar and soil nutrients. Oecologia 89:372–382PubMed
Vogel A, Scherer-Lorenzen M, Weigelt A (2012) Grassland resistance and resilience after drought depends on management intensity and species richness. PLoS One 7:e36992PubMedPubMedCentral
Volk M, Bungener P, Contat F, Montani M, Fuhrer J (2006) Grassland yield declined by a quarter in 5 years of free-air ozone fumigation. Glob Change Biol 12:74–83
Walker BH, Steffen WL, Canadell J, Ingram JSI (1999) The terrestrial biosphere and global change: implications for natural and managed ecosystems, Synthesis volume. Cambridge University Press, Cambridge
Wardle DA, Zackrissen O (2005) Effects of species and functional group loss on island ecosystem properties. Nature 435:806–810PubMed
Weigelt A, Weisser WW, Buchmann N, Scherer-Lorenzen M (2009) Biodiversity for multifunctional grasslands: equal productivity in high-diversity low-input and low-diversity high-input systems. Biogeosciences 6:1695–1706
Wolf B, Merbold L, Decock C, Tuzson B, Harris E, Six J, Emmenegger L, Mohn J (2015) First on-line isotopic characterization of N₂O emitted from intensively managed grassland. Biogeosciences 12:2517–2531
Wright RF (1989) RAIN project—role of organic-acids in moderating pH change following reduction in acid deposition. Water Air Soil Pollut 46:251–259
Wright AJ, Ebeling A, de Kroon H, Roscher C, Weigelt A, Buchmann N, Buchmann T, Fischer C, Hacker N, Hildebrandt A, Leimer S, Mommer L, Oelmann Y, Scheu S, Steinauer K, Strecker T, Weisser W, Wilcke W, Eisenhauer N (2015) Flooding disturbances increase resource availability and productivity, but reduce stability in diverse plant communities. Nat Commun 6:6092PubMed

Prev Next