Drivers of forest dynamics: Joint effects of climate and competition
Full text
Drivers of forest dynamics: Joint effects of climate and competition Dissertation zur Erlangung des akademischen Grades Dr. rer. nat. vorgelegt der Fakultät für Biologie, Chemie und Geowissenschaften der Universität Bayreuth von Frau Dipl. Geoökologin Klara Dolos geboren am 03.03.1982 in Stuttgart
Die vorliegende Arbeit von Frau Dipl.-Geoökol. Klara Dolos, geb. am 03.03.1982 in Stuttgart, wurde in der Zeit von Mai 2009 bis April 2013 unter der Betreuung von Herrn Prof. Dr. Björn Reineking an der Juniorprofessur Biogeographische Modellierung der Universität Bayreuth angefertigt. Vollständiger Abdruck der von der Fakultät für Biologie, Chemie und Geowissenschaften der Universität Bayreuth genehmigten Dissertation zur Erlangung des akademischen Grades Doktor der Naturwissenschaften (Dr. rer. Nat). Dissertation eingereicht am: 25.4. 2013 Zulassung durch die Prüfungskommission: 22.5.2013 Wissenschaftliches Kolloquium: 31.10.2013 Amtierender Dekan: Prof. Dr. Rhett Kempe Prüfungsausschuss: Prof. Dr. B. Reineking (Erstgutachter) Prof. Dr. B. Engelbrecht (Zweitgutachterin) Prof. Dr. M. Hauhs (Vorsitz) Prof. Dr. B. Huwe PD Dr. G Aas 1
Wer immer nur tut, was er schon kann, bleibt immer nur dass, was er schon ist. Henry Ford
Table of contents Abstract........................................................................................................II Zusammenfassung....................................................................................IV Introduction..................................................................................................1 Background.............................................................................................1 Natural forest dynamics..........................................................................3 Demographic processes.....................................................................4 Forest disturbances............................................................................8 Forest succession...............................................................................9 Climatic gradients..................................................................................10 Spatial climatic gradients..................................................................11 Climate change.................................................................................12 Forest modelling....................................................................................13 Statistical models..............................................................................14 Simulation models............................................................................15 Mathematical models........................................................................17 Limits.................................................................................................17 Conclusions...........................................................................................19 References............................................................................................20 Manuscripts...............................................................................................33 Summary of the following manuscripts.................................................33 Manuscript 1: Long-term vegetation dynamics in New Zealand...........35 Manuscript 2: Climatic turning point for beech and oak........................69 Manuscript 3: Symmetric and asymmetric competition......................103 Manuscript 4: Ecotype mixing as climate change adaptation.............133 Manuscript 5: Disturbance interactions...............................................157 Acknowledgements.................................................................................190 Supplement.............................................................................................191 List of manuscripts and specification of own contribution..................191 Presentations on this research...........................................................193 Non-refereed periodicals.....................................................................194 Further publications............................................................................194 Erklärung.............................................................................................195 I
Abstract The present dissertation thesis addresses different aspects of forest dynamics and possible changes due to climate change. Various modelling approaches are used to explore joint effects of climate and competition on forest dynamics with a focus on temperate forests. Motivated by scientific interest, this thesis is aimed at contributing to the establishment of fundamental knowledge for proper ecosystem management. Each of the research projects of this thesis explores a facet of forest dynamics. It appears that for projections of forest dynamics under climate change particularly, it is critical to consider competition among trees. In the first study, the joint effect of climate and competition on forest dynamics in a mountain forest of New Zealand was investigated. The landscape simulation model LandClim was calibrated based on empirical data and applied to reproduce a 1700 years forest succession under stationary climate at the slope of Mt. Hauhungatahi, North Island. Although designed for European temperate forests, LandClim was capable of simulating NZ´s forest dynamics. Under non-stationary climate, forests likely remain in disequilibrium with climate for some time due to the longevity of trees and competitive prevention of establishment. This aspect was investigated in a mixed beech-oak forest in Germany, using LandClim and the forest gap model SILVA in a cooperation study. Furthermore, a possible ‘climatic turning point’ was investigated, the point at which species dominances change due to changes in competitiveness caused by climate change. Both models projected a potential climatic turning point at a mean annual temperature of 11-12 °C and precipitation sum of 500-530 mm. However, the change of species composition in existing mixed stands was much slower since the turning point also depended on inherited stand structure. Based on these projections the promotion of oak at dry sites seems advisable due to its superior resistance and resilience to drought. The applied simulation models consider joint effects of climate and competition but no changes in species sensitivity to competition along climatic gradients. The Spanish National Forest Inventories provided a solid basis to develop a statistical model for the influence of climate and competition on tree growth. The results indicated that in Mediterranean forests the effect of competition increases with aridity potentially resulting in an additional disadvantage for drought sensitive oaks compared to pines under climate change. II
Under the prerequisite that forest dynamics will be affected by climate change, different forest management strategies on adaptation are currently discussed. One option is the promotion of tree species that are better adapted to anticipated future climates (such as oak instead of beech in Germany) and also the introduction of non-local ecotypes of local species. This increase in biodiversity intuitively appears promising because it is in line with the insurance hypothesis and the portfolio effect theory. Within this thesis the effect of ecotype mixing as an increase of within stand diversity under consideration of self-thinning was assessed. It was shown that ecotype mixing in forest stands might lower the risk of yield losses and at the same time might exempt the portfolio effect from its drawback of lower chances for high yields. Climate not only affects demographic rates of tree species but also all other ecosystem components. Disturbances are an important component of forest dynamics because they initiate successions and thereby influence species coexistence. Climate will alter disturbance regimes not only directly but also due to interactions among disturbances, for example an increased risk of insect outbreaks due to weak tree defence caused by severe drought stress. Most disturbance interactions have been observed to be positive, implying that increases of disturbances in quality and quantity due to climate change will be amplified. Furthermore, systems containing positive feedback loops are considered to be mostly unstable, which would result in forest collapse. A theoretical study on disturbance interactions showed why positive feedback loops of disturbances do not necessarily lead to a forest collapse. Disturbance interactions might cause only a minor part of disturbances, whereas direct changes due to climate change are of much higher importance. The described studies reflect the diversity of the research field forest dynamics and innovative ecological methodology. Nevertheless, the present thesis is not an exhaustive discussion of drivers of forest dynamics under climate change. Forest dynamics and its drivers provide a range of open research questions posing a challenge for fundamental an applied research of high relevance for society. III
Zusammenfassung Diese Dissertation befasst sich mit Walddynamik und möglichen Veränderungen in Folge des Klimawandels. Verschiedene Modellierungsansätze werden verwendet, um den Einfluss von Klima und Konkurrenz auf die Walddynamik zu untersuchen. Motiviert durch wissenschaftliche Neugier soll diese Arbeit auch dazu beitragen, fundiertes Wissen als Basis für nachhaltiges Forstmanagement zu schaffen. Jede der hier vorgestellten Studien untersucht einen Aspekt der Walddynamik, mit einem Schwerpunkt auf temperaten Wäldern. Es erweist sich, dass vor allem im Hinblick auf Projektionen zukünftiger Wälder unter Klimawandel Konkurrenz berücksichtigt werden muss. In der ersten hier vorgestellten Studie wurde der gemeinsame Einfluss von Klima und Konkurrenz entlang eines Höhentransektes in Neuseeland untersucht. Das Landschaftsmodell LandClim wurde dazu kalibriert und 1700 Jahre Waldsukzession am Hang des Mt. Hauhungatahi (Nordinsel, Neuseelands) unter stationärem Klima simuliert. Obwohl das Simulationsmodell LandClim für europäische, temperate Wälder entwickelt wurde, war es mit nur geringfügigen Anpassungen möglich, die Walddynamik in Neuseeland zu simulieren. Unter sich veränderndem Klima bleiben Wälder wegen ihrer Langlebigkeit und durch Konkurrenzausschluss von Etablierung für mehrere Generationen im Ungleichgewicht mit dem Klima. Dieser Aspekt wurde in einem deutschen Buchen-Eichen Mischwald unter Verwendung von LandClim und SILVA untersucht. Es wurde nach einem „Klimatischen Wendepunkt“ (climatic turning point, CTP) gesucht, bei dem sich die Dominanz von Buche und Eiche umkehrt. Beide Modelle prognostizierten einen CTP bei Jahresmitteltemperaturen von 11-12 °C und Jahresniederschlägen von 500-530 mm. Ein Artwechsel in bereits bestehenden Beständen fand jedoch nicht sofort statt. Die Studie zeigte, dass es schon heute sinnvoll erscheint auf relativ trockenen Standorten die Eiche gegenüber der Buche zu fördern. Die angewendeten Simulationsmodelle berücksichtigen den gemeinsamen Einfluss von Klima und Konkurrenz auf die Walddynamik, nicht aber eine Veränderung des Einflusses von Konkurrenz entlang eines Klimagradienten. Die spanischen Nationalen Waldinventuren boten die Möglichkeit ein statistisches Modell für den Einfluss von Klima und Konkurrenz auf das Baumwachstum zu entwickeln. Die Ergebnisse zeigten, dass die Bedeutung von Konkurrenz mit zunehmender Aridität größer wird. Dies könnte zu einem zusätzlichen Nachteil für IV
schattentolerante aber dürre-sensitive Arten wie die (sub-) mediterranen Eichen gegenüber Kiefern unter Klimawandel bedeuten. Angesichts des Klimawandels werden aktuell verschiedene Möglichkeiten des Managements diskutiert. Eine vielversprechende Möglichkeit ist die Förderung und auch eine Neu-Einführung von Arten und Ökotypen, die besser an zukünftige Bedingungen angepasst sind. Diese Steigerung der Biodiversität erscheint vielversprechend, da sie im Einklang mit der „Versicherungs-Hypothese“ (engl. insurance hypothesis) und der Portfolio Effekt Theorie ist. In einer Studie konnte gezeigt werden, dass „Ecotype Mixing“, d.h. der Anbau von verschiedenen Ökotypen der selben Art in einem Bestand, das Risiko von Verlusten unter unsicherer Zukunft vermindert. Dabei war der Portfolio Effekt sogar von seinem Nachteil einer geringeren Wahrscheinlichkeit hoher Erträge befreit. „Ecotype Mixing“ erscheint daher eine aussichtsreiche Möglichkeit zur Anpassung an den Klimawandel zu sein. Das Klima beeinflusst nicht nur die demographischen Prozesse der Arten sondern auch alle anderen Ökosystemkomponenten. Störungen machen einen wichtigen Teil der Walddynamik aus, weil sie Sukzessionen initiieren. Dadurch beeinflussen sie die Koexistenz von Arten. Der Klimawandel verändert Störungsregime aber nicht nur direkt sondern auch indirekt, zum Beispiel durch ein erhöhtes Borkenkäfer-Risiko durch vermehrte Dürre und eine dadurch verringerte Widerstandsfähigkeit der Bäume. Die meisten Studien finden positive Interaktionen zwischen einzelnen Störungen, so dass ein Anstieg der Qualität und Quantität von Störungen durch den Klimawandel noch verstärkt würde. Außerdem sind Systeme, die positive Rückkoppelungen enthalten, sehr oft instabil, was zum Zusammenbruch der Wälder führen könnte. Eine theoretische Studie über Störungsinteraktionen zeigte, warum positive Rückkoppelungen zwischen Störungen nicht zum vollständigen Rückgang der Wälder führen müssen. Außerdem zeigte die Studie, dass Störungsinteraktionen möglicherweise nur einen geringen Anteil an Störungsregimen bedingen und Veränderungen im Zuge des Klimawandels weitaus bedeutender sind. Die vorgestellten Studien spiegeln die Vielfalt des Forschungsfeldes „Walddynamik“ und der ökologischen Methodik wieder, es was aber unmöglich das Thema umfassend im Rahmen dieser Dissertation abzuhandeln. Es bleiben daher eine Vielzahl von offenen Fragen zu Walddynamik und den beteiligten Prozessen, deren Erforschung eine Herausforderung darstellt und von großer Bedeutung für die Gesellschaft ist. V
Introduction Introduction Background Facilitation of sustainable forest management is a crucial task to protect biodiversity, to reduce carbon emissions from deforestation and forest degradation while preserving the socio-economic value. Trees as the key organisms of forests provide shelter and food for animals such as large herbivores and smaller mammals, birds, insects, as well as for understorey plants, epiphytes, mosses, fungi and, notably, to a rich biodiversity of soil organisms. Estimates of global tree species richness range from 60,000 (Grandtner 2005) to 100,000 taxa (Oldfield et al. 1998). Besides the intrinsic value of biodiversity (Lindenmayer et al. 2000) and its role for ecological stability, forests provide very important goods and services to society such as watershed and air quality protection, soil erosion control, food and timber production, and recreational and cultural functions (Schroter 2005, Sunderlin et al. 2008, Turner and Daily 2008). Current climate change is anticipated to affect forest productivity, composition and distribution considerably (Allen et al. 2010, Köhl et al. 2010, Hanewinkel et al. 2012). Besides alterations in mean climate, increasing temporal variability and more frequent occurrences of extreme weather events such as droughts pose challenges for ecosystem management (Jentsch 2007, Jentsch and Beierkuhnlein 2008). Notably, projected impacts of climate change on forests differ among bioclimatical regions (Rivas-Martínez et al. 2004, Scholze 2006). Temperate and boreal forests, for example, are less vulnerable in regard to productivity losses than Mediterranean forest ecosystems, where growth is already water limited (Lindner et al. 2010). The management of long-lived ecosystems such as forests has to deal with uncertain future growing conditions while long term decisions have to be made (Hildebrandt and Knoke 2011). Particularly in the face of climate change, management decisions that are only based on experience but lack profound understanding cannot cope with dynamic changes of ecosystem processes provisioning biotic interactions (Korzukhin et al. 1996). This is based on the assumption that future conditions have no equivalent in the past (Williams and Jackson 2007). Under such novel conditions forest community responses cannot simply be pro- 1
Natural forest dynamics Another source of data on demographic rates are experimental studies (Underwood 2009). In manipulative experiments environmental variables are controlled while the target variable, e.g. growth, is measured (Manuscript 4). This approach facilitates separating the effects of variables that are largely correlated in the field (such as temperature and precipitation, stand density and precipitation; Manuscript 3). Additionally, by manipulative experiments the responses of an organism or ecosystem to conditions of interest can be assessed systematically through manipulative experiments. Consequently, experiments are very common in climate change research (Bart 2006, Jentsch et al. 2007, Pretzsch et al. 2010, Kreyling et al. 2012). However, slow processes such as tree growth and regular mortality result in the need for long-lasting experiments (Benson et al. 1992, Pretzsch et al. 2010). Since it is not always feasible to wait for the entire life-span of a tree to get results, alternative model organisms are often used. To assess the growth reaction of tree species and ecotypes to different climatic conditions, saplings are commonly used and findings are transferred to mature trees (Thiel et al. 2012, Manuscript 4). Aside from the problem of transferability, the need for replicates to enable hypothesis testing poses a main challenge (Oksanen 2001). Forest disturbances Disturbances constitute an important component of forest dynamics (Jentsch 2007). Their occurrence is rather sporadic or can re-initiate forest successions at broader scales. On a global scale, the main non-anthropogenic forest disturbances are wind storms, wildfires and insect outbreaks (Dale et al. 2000, Dale et al. 2001, Manuscript 5). Further examples of natural disturbances are herbivores, extreme weather events such as drought (Ayres and Lombardero 2000, White and Jentsch 2001, Weisberg and Bugmann 2003) and extreme disturbances such as volcanic eruptions affecting entire landscapes (‘large infrequent disturbances’ sensu Turner et al. 1998, Manuscript 1). Generally, in forest ecology disturbance is an event that involves the removal of biomass or causes death of individuals (Grime 1977, Sousa 1984) and can be characterized by its frequency, size, severity, intensity and residuals (White and Pickett 1985, Turner et al. 1998). Temporal scaling of disturbances, i.e. their frequency, has to be interpreted with regard to the generation time of the population of interest, i.e. the life span of trees in the case of forests. The same applies to the 8
Natural forest dynamics spatial scaling which is relative to the organism size or its range of activity, for trees dispersal distances. In forests, severity usually includes not only biomass removal and tree mortality but also the disruption of ecosystem structure: Erosion due to exposure of mineral soil and loss of soil nutrients, death of advance regeneration and seed banks (Peterson and Pickett 1995). The sum of all disturbances affecting an ecosystem is called its disturbance regime. Disturbances are considered to affect biodiversity and species coexistence (Nakashizuka 2001, Papaik and Canham 2006). According to the predictions of the intermediate disturbance hypothesis (Connell 1978, Huston 1979), diversity is high at intermediate disturbance levels due to the coexistence of colonizers and competitors (Hubbell 1999, Stueve et al. 2011, Manuscript 1). The underlying mechanism is biomass removal from the community. Thereby, competitive species are prevented from rapidly occupy the entire area (Manuscript 2; but see also Tilman 1994 for the spatial competition hypothesis). As a result, the community remains in an early to intermediate successional phase. In the climate change context, disturbances can be considered as catalysts of vegetation changes based on their effect to re-initiate successions with unknown successional pathways (Sykes and Prentice 1996, Soja et al. 2007, Manuscript 2). Disturbance regimes will likely alter due to climate change (Dale et al. 2000, Dale et al. 2001), either because they are directly related to climate such as drought or wind storms (Jentsch et al. 2007) or because they are indirectly influenced by climate such as wildfires by ‘fire weather’, and insect outbreaks due to temperature control of the organisms life cycles (Wermelinger 2004, Bigler et al. 2005, Carnicer et al. 2011). Furthermore, disturbances can exert a feedback on vegetation properties such as fuel load for fires and stand age and composition for insects such as bark beetles (Veblen et al. 1994, Cumming 2001). Disturbances also interact among each other and thereby complicate predictions of disturbance regimes under changed environmental conditions (White and Jentsch 2001, Buma and Wessman 2011, Manuscript 5). Forest succession Forest succession is the process of change in species structure and composition of a forest community over time (Finegan 1984, Pickett et al. 1987). Succession pathways differ between climates, soils and regional species pools (Kupfer and 9
Natural forest dynamics Runkle 1996, Manuscripts 1 & 2). Species replacements during forest succession are primarily influenced by changes in light conditions (Finegan 1984, Franklin et al. 2002). Furthermore, soil (i.e. nutrient cycling) and other factors change with time (Guariguata and Ostertag 2001, Burga et al. 2010). In an ideal case, succession begins uninfluenced by pre-existing communities (primary succession). First colonizers can usually be described as light demanding pioneer species (such as Betula pendula, Pinus sylvestris, Sorbus aucuparia in temperate forests) with high dispersal potential and fast growth rates (Grime 1977). These early successional species will be replaced over time by competitive species that are more shade tolerant, growing at slower pace and have longer live span (e.g. Fagus sylvatica). Until recently, a stable climax community was considered to be the end point of succession (Mueller-Dombois and Ellenberg 1974). Over the course of time, though, a stand is likely to be affected by disturbance events. Currently, the climax theory is supplemented by the acknowledgement that ecosystems change continuously and that in many cases a climax state is never attained due to regular small scale disturbances (Pickett 1980, van der Maarel et al. 1995, Yamamoto 2000). Climatic gradients Climatic gradients can be of spatial (latitude, continentality, altitude) and temporal nature (‘climate change’). Tree species performance (e.g. rates of demographic processes) and thereby community assembly change along climatic gradients is mainly due to differences in energy and water supply (O'Brien et al. 2000). There is a fundamental difference between spatial and temporal gradients for sessile organisms such as plants, particularly for long-lived organisms. Under rapid (temporal) climate change individual trees might be subject to different climates. Thereby, current climate change mixes effects of a climatic gradient with forest succession (as in Manuscript 2) whereas effects of spatial climatic gradients are mixed with alterations of other site properties such as soil properties (as in Manuscripts 1 & 3). Relative species competitiveness is determined by performance and therefore changes with climate (despite e.g. nutrient limited sites). Community responses to climate are hence a combination of single species responses and interactions among species. A number of studies shows that the nature of competition changes along ecological gradients (Kunstler et al. 2011, Zang et al. 2012, Manuscript 3). 10
Climatic gradients Changes in species competitiveness cause rank reversals in species performance and thereby result in non-linear changes in community structure (Sánchez-Gómez et al. 2008, Gómez-Aparicio et al. 2011, Scharnweber et al. 2011, Manuscript 2). Spatial climatic gradients Global circulation causes global climatic zones (Flohn 1957, Salmond and Smith 1996). Generally, higher latitudes comprise cooler climates since the amount of solar radiation gets dispersed over a larger area than in the central latitudes. This pattern is mainly modified by planetary wind systems transporting latent energy. Moreover, cool or warm ocean currents transport energy from lower to higher latitudes (e.g. the warm Gulf Stream). Smaller currents influence temperatures close to the coast (e.g. the cold Canary Stream). Precipitation is strongly influenced by the global and local wind systems, land-water distributions and relief (Street-Perrott et al. 1983, Trenberth 1999). At the continental scale, latitudinal climatic zones are modified by the effect of continentality (distance from the sea). Regions close to large water bodies are usually moister and comprise a lower temperature variability. In contrast, regions within land surfaces have larger temperature amplitudes and less precipitation. This difference is a consequence of much lower effective heat capacities of land surfaces as well as of their reduced evaporation rates compared to large water bodies. Altitudinal gradients provide the opportunity to assess vegetation patterns within small areas but with large environmental differences (Beals 1969, Jump et al. 2009). Similar to latitudinal gradients, species distribution along altitudinal gradients is primarily controlled by climate, particularly the upper limits (Wardle 1964, Druitt et al. 1990, Manuscript 1). Lowland environments are usually warmer and drier and maintain a higher site productivity. Temperature falls with increasing altitude due to adiabatic cooling and relief rainfall usually leads to higher precipitation. In addition to the climatic gradient other factors such as soil properties change with altitude and should to be considered in the interpretation of species and community zonation (Brzeziecki et al. 1993, Burga et al. 2010). 11
Climatic gradients Climate change Major climatic changes have occurred several times throughout the past millennia (Petit et al. 1999). The last strong climatic change took place c. 10,000 years ago after the Quaternary glaciation cycles. In Europe, tree species expanded their distribution ranges from their refugia northwards (Taberlet et al. 1998). Contemporary forest communities are considered to be the result of this post-glacial colonization which is most likely still in progress. Current climate change can be observed best in mountain regions in terms of upward shifts of species ranges since the dispersal distances of species coincide with the rate of climate change (Cannone et al. 2008, Harsch et al. 2009, Jump et al. 2009). In contrast, horizontal species range shifts are dispersal limited, due to large distances and landscape fragmentation. The rate of previous climatic changes during the Quaternary was smaller than the rate of current anthropogenic climate change. Migration rates of tree species have been estimated from 50 to 2000 m per year (Davis 1983 for Fagus grandifolia and Dennis 1993 for Picea spp.). Many species have the potential in terms of dispersal rate to keep up with moderate climatic changes and are likely to shift their distribution ranges in the future. Moreover, given sufficient time, evolutionary adaptation might occur, i.e. genetic changes generating better adapted ecotypes or even new species. In contrast, it is unclear if current fast-paced climatic change exceeds the limits of adaptability for some species possible resultung in local and global extinctions (Jump et al. 2009; refere to Kreyling et al. 2011 for assisted migration). Mountains and their unique biota are disproportionately vulnerable to climate change (Theurillat and Guisan 2001, Engler et al. 2011, Dullinger et al. 2012). On major reason is that species cannot escape the upward shift of climatic zones once the mountain summit is reached. New sites with suitable environmental conditions for mountain species are often very distant and outside the range of natural dispersal. Current climate change is faster than forest succession due to the longevity of trees. Thereby, forests will likely remain in non-equilibrium with climate. In forest stands, effects of climate change might be observable with a time lag in the order of life-span of a tree (Kienast and Kräuchi 1991, Sykes and Prentice 1996, Manuscript 1). 12
Forest modelling Forest modelling Ecological models serve as tools to extend conceptual ideas and learn about interrelationships in ecosystems. From an ecologists view models can be used for •testing hypotheses about the functioning of ecosystems (Manuscripts 1-3), •exploring system behaviour under different (climatic) conditions (Manuscripts 2, 5), •investigating scenarios which are impossible to test in real ecosystems (Manuscripts 1, 2, 4, 5), •supplementing and motivating field experiments and further data collection (Manuscripts 4 & 5), and •identifying knowledge gaps and developing new hypothesis (Manuscripts 4 & 5) (e.g. Caswell 1988, Wissel 1992, Dislich 2011). Like every scientific endeavour, model building follows the basic principles of science (Grimm 1994, Jackson et al. 2000). Observations of the natural world should be foundation of all ecological research, including model building. Additionally, ecological knowledge and system understanding are the basis for developing new ecological theory and hypotheses (preferably but not exclusively testable ones; Caswell 1988). Yet, understanding systems and processes is crucial, an incomplete understanding of the system should not prevent model building (Starfield 1997). For model building there is need for abstraction from real ecosystems and reduction of complexity with respect to the formulated hypotheses (Wissel 1989, Grimm 1999). Structures and processes of interest must be defined and decisions about spatial and temporal scales as well as the type of model must be taken. By means of combining models with empirical data, the system behaviour can be explored and hypotheses can be tested. Finally, as modelling is an iterative process, the results must be evaluated. Modifications of the model and its underlying theory should be discussed and new data surveys or experiments might be designed. 13
Forest modelling In the following sections three main types of models are introduced which can be used to explore forest dynamics and its drivers (Manuscripts 1-5). Each type of model comprises different advantages and limitations essential to acknowledge. Statistical models In natural resource planning, models are typically used to project how resource dynamics will develop in the future and to guide decision making. A traditional method to project forest growth and yield is the use of regional and site-specific yield tables which interpolate past observations of stand growth under given climatic, soil and management conditions (e.g. Assmann and Franz 1972). Contemporary statistical models (empirical models) of tree species distributions, site suitability, growth and productivity are more flexible than yield tables and are fitted to much larger datasets. There are many different types of statistical models but all of them base on the principle of drawing conclusions from data that is subject to random variation. Statistical models aid in the exploration of the effect of variables on observed patterns and thus give phenomenological descriptions of vegetation responses (Wissel 1992, Dormann et al. 2012). Prediction accuracy is mostly an implicit objective of statistical modelling, particularly when used as management support. Prominent examples of applications are species distribution models (SDMs; Araújo and Guisan 2006, Falk and Mellert 2011), models on demographic processes of trees (Kunstler et al. 2011, Ruiz-Benito et al. 2012, Ruiz-Benito et al. 2013, Manuscript 3) and also statistical down-scaling of global circulation models in climate change research (e.g. WETTREG; Enke et al. 2005a, Enke et al. 2005b). Since statistical models often perform better than process-based models in terms of predictions, it is advisable to use them for management questions. However, management based on SDMs ignores abundant information available from forest inventories (e.g. Kölling 2007, Falk and Mellert 2011). Species range limits are without doubt strongly influenced by climate and soil properties, but other factors might be of similar importance (Sexton et al. 2009), depending on the considered scale (Pearson and Dawson 2003). Firstly, species presence in Europe is strongly influenced by management resulting in a logical cycle when calculating site suitability based on species presences and absences (Hannah et al. 14
Forest modelling 1995). Secondly, dispersal limitation might limit distributions more than abiotic site suitability (Svenning and Skov 2007). Thirdly, distribution is influenced by biotic interactions such as competition (which might be reduced by management; e.g. Pretzsch et al. 2013). Therefore, species niches and site suitability inferred from current range limits might not represent the actual physiological potential of a species. The analysis of demographic processes might help to overcome these limitations. In any case, statistical models do not readily allow for insights into processes leading to the observed pattern as they follow the general concept of correlation. Consequently, effect and response might be mixed and false correlations might lead to misinterpretations. Moreover, correlative approaches only work well in constant environments but not under non-analogous conditions triggered by climate change (Bossel 1991, Williams and Jackson 2007). Simulation models Simulation models have advantages over statistical models when considering projections outside known environments and consequently gain importance in forest management applications (Pretzsch et al. 2008, Muys et al. 2010). Forest simulation models are based on the current understanding of biomass, energy, water, nutrients and competition relations as other biotic interactions. Ecosystem structure and processes can be implemented based on mechanistic understanding (Tilman 1987, Bossel 1991). Single-tree models with an emphasize on either morphology or physiology are very detailed models. Morphological models aim at simulating the spatial structure of a plant (Kurth 1994) while physiological models (process-based models) simplify the spatial configuration but start with biochemical representations of photosynthesis and respiration controlled by resource availability in high temporal resolution (Grote and Pretzsch 2002). Furthermore, there have also been efforts to combine the advantages of both model approaches (Perttunen 1996). As the scale of interest becomes coarser and larger systems are considered, smaller components are generally aggregated to maintain a manageable complexity within models (Levin 1992, Rastetter et al. 1992, Elkin et al. 2012). Forest gap models are at intermediate spatial and temporal scales and simulate forest stands for one generation or rotation period (Botkin et al. 1972, Shugart and West 15
Forest modelling 1981, Bugmann 2001). With this approach physiological processes are aggregated and empirical (correlative, statistical) functions are used to relate demographic processes to the environment. Competition among neighbouring trees is considered morphologically simplified. Gap models have a high potential for application in management questions as they can be used to explore the stand-level implications of management regimes, changes of atmospheric CO2 concentrations, nitrogen deposition, or climate change (Mäkelä et al. 2000). The forest growth simulator SILVA, for example, was primarily developed as a management tool, but can also be applied to ecological questions concerning climate change (Pretzsch et al. 2002, Muys et al. 2010, Manuscript 2). LandClim is an example of a forest model simulating climate and competitiondriven forest dynamics and disturbance regimes at a landscape-scale (Schumacher et al. 2004, Schumacher et al. 2006, Manuscripts 1 & 2). Forest simulations at such larger scales need further aggregation of ecosystem processes compared to gap models. LandClim uses a grid based approach with a spatial resolution of c. 25 x 25 m per grid cell being detailed enough to capture local variation in soil conditions and topography. This model simulates species-level forest dynamics by tracking species age cohorts (not individual trees) at 10-year time steps under natural disturbances (e.g. Colombaroli et al. 2010, Henne et al. 2012). Similar to LandClim, LANDIS is a spatially explicit, stochastic, raster-based landscape model which facilitates the study of the effects of natural and anthropogenic disturbances, vegetational succession, management strategies and their interactive effects on forest landscapes (Mladenoff et al. 1999). LANDIS is a powerful tool for evaluating alternative forest management strategies at a landscape-scale (He 1999, Mehta et al. 2004, Xu et al. 2009, Yao et al. 2012). Dynamic Global Vegetation Models (DGVMs) simulate the global distribution of forests and other vegetation types under historic, present and simulated future climates (Prentice et al. 2007, Purves and Pacala 2008). Accurate representation of the structure and processes of natural systems is necessary for reliable ecological models. Yet, for DGVMs generalization of unneeded detail is required to increase model efficiency. In DGVMs, biomes, vegetation types or plant functional types are reduced to their biomass or net primary productivity (Peng 2000, Bonan et al. 2003, Sitch et al. 2008). Applications of such models include global risk assessments for ecosystems under projected climate change (Scholze 2006), the effect of 16
Forest modelling forest carbon uptake on climate (Purves and Pacala 2008) and the role of agriculture within the global climate-vegetation system (Bondeau et al. 2007). Mathematical models Complex simulation models of ecological systems are as a rule incomprehensible and the behaviour of the model cannot be fully explored. In contrast, simplistic mathematical models (conceptual/strategic mathematical models; May 1976, Grimm 1994) offer analytically tractable solutions and the possibility to examine the underlying model properties. This allows for understanding the parts of the model which determine the resulting pattern. Mathematical models thus ‘sacrifice detail for generality’ and thereby serve to demonstrate the ‘consequences of what we believe to be true’ (Levins 1966, Grimm 1994, Manuscript 5). Models of population dynamics, the main field of mathematical modelling in ecology, are often expressed in terms of differential or difference equations, which describe how system components change over time. The most widely known mathematical models stem form the Lotka-Volterra equations describing an ecological predator-prey (or parasite-host) system (e.g. Berryman 1992). The model is largely phenomenological, combining simple ‘laws’ of exponential growth of populations not limited by resources (Malthus 1798) and of mass action kinetics borrowed from chemistry (Murray 2002). The logistic model of population growth integrates the concept of negative density dependence. As populations grow, resources decline, so that fitness of any population member is negatively correlated to population density (cf. Berryman 1999). This principle was transferred to tree growth as growth-size allometry (Zeide 1993). Another example is the principle of competitive exclusion which was assessed by another descendent of the Lotka-Volterra model (Cushing et al. 2004). Limits As modelling has become a frequently used tool in ecology, it is important to be aware of problems which models cannot resolve. Models are imperfect representations of real ecosystems, by definition and purpose, and due to limited system understanding and computation power. They always contain errors in assumption, formulation and parametrization (Grimm 1999). Levins (1966) points out that 17
References Franklin, J. F., Spies, T. A., van Pelt, R., Carey, A. B., Thornburgh, D. A., Berg, D. R., Lindenmayer, D. B., Harmon, M. E., Keeton, W. S., Shaw, D. C., Bible, K. and Chen, J. 2002. Disturbances and structural development of natural forest ecosystems with silvicultural implications, using Douglas-fir forests as an example. – Forest Ecology and Management 155: 399- 423. García Valdés, R., Zavala, M. A., Araújo, M. B. and Purves, D. W. 2013. Chasing a moving target: ‐ Projecting climate change induced shifts in non equilibrial tree species distributions. – Journal ‐ ‐ of Ecology 101: 441-453. Gitlin, A. R., Sthultz, C. M., Bowker, M. A., Stumpf, A., Paxton, K. L., Kennedy, K., Muñoz, A., Bailey, J. K. and Whitham, T. G. 2006. Mortality gradients within and among dominant plant populations as barometers of ecosystem change during extreme drought. – Conservation Biology 20: 1477-1486. Gómez-Aparicio, L., García-Valdés, R., Ruiz-Benito, P. and Zavala, M. A. 2011. Disentangling the relative importance of climate, size and competition on tree growth in Iberian forests: Implications for forest management under global change. – Global Change Biology 17: 2400-2414. Grandtner, M. M. 2005. Elsevier's Dictionary of Trees, 1. North America – Elsevier. Grime, J. P. 1977. Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary theory. – The American Naturalist 111: 1169. Grimm, V. 1994. Mathematical models and understanding in ecology. – Ecological Modelling 75- 76: 641-651. Grimm, V. 1999. Ten years of individual-based modelling in ecology: What have we learned and what could we learn in the future? – Ecological Modelling 115: 129-148. Grimm, V. and Railsback, S. 2012. Pattern-oriented modelling: A ‘multi-scope’ for predictive systems ecology. – Philosophical Transactions of the Royal Society B: Biological Sciences 367: 298 310.‐ Grote, R. and Pretzsch, H. 2002. A model for individual tree development based on physiological processes. – Plant Biology 4: 167-180. Guariguata, M. R. and Ostertag, R. 2001. Neotropical secondary forest succession: Changes in structural and functional characteristics. – Forest Ecology and Management 148: 185-206. Hanewinkel, M., Cullmann, D. A., Schelhaas, M.-J., Nabuurs, G.-J. and Zimmermann, N. E. 2012. Climate change may cause severe loss in the economic value of European forest land. – Nature Climate Change 3: 203-207. Hannah, L., Carr, J. L. and Lankerani, A. 1995. Human disturbance and natural habitat: A biome level analysis of a global data set. – Biodiversity and Conservation 4: 128-155. Harper, E. B., Stella, J. C. and Fremier, A. K. 2011. Global sensitivity analysis for complex ecological models: A case study of riparian cottonwood population dynamics. – Ecological Applications 21: 1225-1240. Harsch, M. A., Hulme, P. E., McGlone, M. S. and Duncan, R. P. 2009. Are treelines advancing? A global meta-analysis of treeline response to climate warming. – Ecology Letters 12: 1040‐ 1049. He, H. S. 1999. The effects of seed dispersal on the simulation of long-term forest landscape change. – Ecosystems 2: 308-319. Hebertson, E. G. and Jenkins, M. J. 2008. Climate factors associated with historic spruce beetle (Coleoptera Curculionidae) outbreaks in Utah and Colorado. – Environmental Entomology 37: 281-292. 24
References Henne, P. D., Elkin, C., Colombaroli, D., Samartin, S., Bugmann, H., Heiri, O. and Tinner, W. 2012. Impacts of changing climate and land use on vegetation dynamics in a Mediterranean ecosystem: Insights from paleoecology and dynamic modeling. – Landscape Ecology. Higgins, S. I. and Richardson, D. M. 1999. Predicting plant migration rates in a changing world: The role of long distance dispersal. – The American Naturalist 153: 464-475.‐ Hildebrandt, P. and Knoke, T. 2011. Investment decisions under uncertainty - A methodological review on forest science studies. – Forest Policy and Economics 13: 1-15. Houghton, R. A. 1994. The worldwide extent of land-use change. – BioScience 44: 305-313. Howe, H. F. and Smallwood, J. 1982. Ecology of seed dispersal. – Annual Review of Ecology and Systematics 13: 201-228. Hubbell, S. P. 1999. Light-gap disturbances, recruitment limitation, and tree diversity in a neotropical forest. – Science 283: 554-557. Huston, M. 1979. A general hypothesis of species diversity. – The American Naturalist 113: 81. Huston, M. and Smith, T. 1987. Plant succession: Life history and competition. – The American Naturalist 130: 168. Jackson, L. J., Trebitz, A. S. and Cottingham, K. 2000. An introduction to the practice of ecological modeling. – BioScience 50: 694. Jentsch, A. 2007. The challenge to restore processes in face of nonlinear dynamics? On the crucial role of disturbance regimes. – Restoration Ecology 15: 334-339. Jentsch, A. and Beierkuhnlein, C. 2008. Research frontiers in climate change: Effects of extreme meteorological events on ecosystems. – Comptes Rendus Geoscience 340: 621-628. Jentsch, A., Kreyling, J. and Beierkuhnlein, C. 2007. A new generation of climate-change experiments: Events, not trends. – Frontiers in Ecology and the Environment 5: 365-374. Jump, A. S., Mátyás, C. and Peñuelas, J. 2009. The altitude-for-latitude disparity in the range retractions of woody species. – Trends in Ecology & Evolution 24: 694-701. Keane, R. E., Austin, M., Field, C., Huth, A., Lexer, M. J., Peters, D., Solomon, A. M. and Wyckoff, P. 2001. Tree mortality in gap models: Application to climate change. – Climatic Change 51: 509-540. Kellman, M. and Kading, M. 1992. Facilitation of tree seedling establishment in a sand dune succession. – Journal of Vegetation Science 3: 679-688. Kienast, F. and Kräuchi, N. 1991. Simulated successional characteristics of managed and unmanaged low-elevation forests in Central Europe. – Forest Ecology and Management 42: 49-61 Kobe, R. K., Pacala, S. W., Silander, J. A. and Canham, C. D. 1995. Juvenile tree survivorship as a component of shade tolerance. – Ecological Applications 5: 517. Köhl, M., Hildebrandt, R., Olschofksy, K., Köhler, R., Rötzer, T., Mette, T., Pretzsch, H., Köthke, M., Dieter, M., Abiy, M., Makeschin, F. and Kenter, B. 2010. Combating the effects of climatic change on forests by mitigation strategies. – Carbon Balance and Management 5: 8. Kölling, C. 2007. Klimahüllen für 27 Waldbaumarten. – AFZ-DerWald 23: 1242-1245. Koop, H. 1987. Vegetative reproduction of trees in some European natural forests. – Vegetatio 72: 103-110. Korzukhin, M. D., Ter-Mikaelian, M. T. and Wagner, R. G. 1996. Process versus empirical models: which approach for forest ecosystem management? – Canadian Journal of Forest Research 26: 879-887. Kowarik, I. and Boye, P. 2003. Biologische Invasionen - Neophyten und Neozoen in Mitteleuropa. Mit 76 Tabellen – Ulmer. 25
References Kozłowski, J. 1991. Optimal energy allocation models - An alternative to the concepts of reproductive effort and cost of reproduction. – Acta Oecologica 12: 11-33. Kozłowski, J. 1992. Optimal allocation of resources to growth and reproduction: Implications for age and size at maturity. – Trends in Ecology & Evolution 7: 15-19. Kreyling, J., Bittner, T., Jaeschke, A., Jentsch, A., Jonas Steinbauer, M., Thiel, D. and Beierkuhnlein, C. 2011. Assisted colonization: A question of focal units and recipient localities. – Restoration Ecology 19: 433-440. Kreyling, J., Thiel, D., Simmnacher, K., Willner, E., Jentsch, A. and Beierkuhnlein, C. 2012. Geographic origin and past climatic experience influence the response to late spring frost in four common grass species in Central Europe. – Ecography 35: 268-275. Kunstler, G., Albert, C. H., Courbaud, B., Lavergne, S., Thuiller, W., Vieilledent, G., Zimmermann, N. E. and Coomes, D. A. 2011. Effects of competition on tree radial-growth vary in importance but not in intensity along climatic gradients. – Journal of Ecology 99: 300-312. Kupfer, J. A. and Runkle, J. R. 1996. Early gap successional pathways in a Fagus-Acer. – Journal of Vegetation Science 7: 247-256. Kurth, W. 1994. Morphological models of plant growth: Possibilities and ecological relevance. – Ecological Modelling 75-76: 299-308. Leakey, R. R. 1985. The capacity for vegetative propagation in trees. – In: Cannell, M. G. R. and Jackson, J. E. (eds.), Attributes of trees as crop plants. Institute of Terrestrial Ecology, Natural Environment Research Council, pp. 110–133. Levin, S. A. 1992. The problem of pattern and scale in ecology: The Robert H. MacArthur Award Lecture. – Ecology 73: 1943. Levins, R. 1966. Strategy of model building in population biology. – American Scientist 54: 421- 431. Lindenmayer, D. B., Margules, C. R. and Botkin, D. B. 2000. Indicators of biodiversity for ecologically sustainable forest management. – Conservation Biology 14: 941-950. Lindner, M., Maroschek, M., Netherer, S., Kremer, A., Barbati, A., Garcia-Gonzalo, J., Seidl, R., Delzon, S., Corona, P., Kolström, M., Lexer, M. J. and Marchetti, M. 2010. Climate change impacts, adaptive capacity, and vulnerability of European forest ecosystems. – Forest Ecology and Management 259: 698-709. Littell, J. S., McKenzie, D., Kerns, B. K., Cushman, S. and Shaw, C. G. 2011. Managing uncertainty in climate-driven ecological models to inform adaptation to climate change. – Ecosphere 2: art102. Loehle, C. 1998. Height growth rate trade-offs determine northern and southern range limits for trees. – Journal of Biogeography 25: 735-742. Lonsdale, W. M. 1990. The self-thinning rule: Dead or alive? – Ecology 71: 1373. Lusk, C. H. 2004. Leaf area and growth of juvenile temperate evergreens in low light: Species of contrasting shade tolerance change rank during ontogeny. – Functional Ecology 18: 820-828. Luterbacher, J. 2004. European seasonal and annual temperature variability, trends, and extremes since 1500. – Science 303: 1499-1503. Mäkelä, A., Landsberg, J., Ek, A. R., Burk, T. E., Ter-Mikaelian, M., Agren, G. I., Oliver, C. D. and Puttonen, P. 2000. Process-based models for forest ecosystem management: Current state of the art and challenges for practical implementation. – Tree Physiology 20: 289-298. 26
References Makler-Pick, V., Gal, G., Gorfine, M., Hipsey, M. R. and Carmel, Y. 2011. Sensitivity analysis for complex ecological models - A new approach. – Environmental Modelling & Software 26: 124-134. Malthus, T. R. 1798. An essay on the principle of population – J. Johnson. Mantgem, P. J. van and Stephenson, N. L. 2007. Apparent climatically induced increase of tree mortality rates in a temperate forest. – Ecology Letters 10: 909-916. Manusch, C., Bugmann, H., Heiri, C. and Wolf, A. 2012. Tree mortality in dynamic vegetation models – A key feature for accurately simulating forest properties. – Ecological Modelling 243: 101-111. Manzano, P. and Malo, J. E. 2006. Extreme long-distance seed dispersal via sheep. – Frontiers in Ecology and the Environment 4: 244-248. Martinelli, N. 2004. Climate from dendrochronology: Latest developments and results. – Global and Planetary Change 40: 129-139. May, R. M. 1976. Simple mathematical models with very complicated dynamics. – Nature 261: 459-467. McDowell, N. G. and Sevanto, S. 2010. The mechanisms of carbon starvation: How, when, or does it even occur at all? – New Phytologist 186: 264-266. McDowell, N. G., White, S. and Pockman, W. T. 2008. Transpiration and stomatal conductance across a steep climate gradient in the southern Rocky Mountains. – Ecohydrology 1: 193-204. Mehta, S., Frelich, L. E., Jones, M. T. and Manolis, J. 2004. Examining the effects of alternative management strategies on landscape-scale forest patterns in northeastern Minnesota using LANDIS. – Ecological Modelling 180: 73-87. Meinardus, C. and Bräuning, A. 2011. Zur Trockenstresstoleranz von Eichen und Buchen. Erholungsreaktionen der Rotbuche und der Traubeneiche nach Dürrejahren. – LWF aktuell 85: 9-11. Meiners, S. 2000. Tree seedling establishment under insect herbivory: Edge effects and interannual variation. – Plant Ecology (formerly Vegetatio) 151: 161-170. Michelot, A., Bréda, N., Damesin, C. and Dufrêne, E. 2012. Differing growth responses to climatic variations and soil water deficits of Fagus sylvatica, Quercus petraea and Pinus sylvestris in a temperate forest. – Forest Ecology and Management 265: 161-171. Mladenoff, D. J. et al. 1999. Design, behavior and application of LANDIS, an object-oriented model of forest landscape disturbance and succession – Cambridge University Press. Mueller-Dombois, D. and Ellenberg, H. 1974. Aims and methods of vegetation ecology – Wiley. Murray, J. D. 2002. Mathematical biology - An introduction – Springer. Muys, B., Hynynen, J., Palahi, M., Lexer, M. J., Fabrika, M., Pretzsch, H., Gillet, F., Briceno, E., Nabuurs, G. J. and Kint, V. 2010. Simulation tools for decision support to adaptive forest management in Europe. – Forest Systems 19: 86-99. Nakashizuka, T. 2001. Species coexistence in temperate, mixed deciduous forests. – Trends in Ecology & Evolution 16: 205-210. Niinemets, Ü. 2006. The controversy over traits conferring shade-tolerance in trees: Ontogenetic changes revisited. – Journal of Ecology 94: 464-470. Noss, R. F. 2001. Beyond Kyoto: Forest management in a time of rapid climate change. – Conservation Biology 15: 578-590. 27
References O'Brien, E. M., Field, R. and Whittaker, R. J. 2000. Climatic gradients in woody plant (tree and shrub) diversity: Water-energy dynamics, residual variation, and topography. – Oikos 89: 588- 600. Oksanen, L. 2001. Logic of experiments in ecology: Is pseudoreplication a pseudoissue? – Oikos 94: 27-38. Oldfield, S., Lusty, C. and MacKinven, A. 1998. The world list of threatened trees – World Conservation Press. Papaik, M. J. and Canham, C. D. 2006. Species resistance and community response to wind disturbance regimes in northern temperate forests. – Journal of Ecology 94: 1011-1026. Pearson, R. G. and Dawson, T. P. 2003. Predicting the impacts of climate change on the distribution of species: Are bioclimate envelope models useful? – Global Ecology and Biogeography 12: 361-371. Peng, C. 2000. From static biogeographical model to dynamic global vegetation model: A global perspective on modelling vegetation dynamics. – Ecological Modelling 135: 33-54. Perttunen, J. 1996. LIGNUM: A tree model based on simple structural units. – Annals of Botany 77: 87-98. Peterson, C. J. and Pickett, S. T. A. 1995. Forest reorganization: A case study in an old-growth forest catastrophic blowdown. – Ecology 76: 763. Peterson, G. D., Cumming, G. S. and Carpenter, S. R. 2003. Scenario planning: A tool for conservation in an uncertain world. – Conservation Biology 17: 358-366. Petit, J. R., Jouzel, J., Raynaud, D., Barkov, N. I., Barnola, J.-M., Basile, I., Bender, M., Chappellaz, J., Davis, M., Delaygue, G., Delmotte, M., Kotlyakov, V. M., Legrand, M., Lipenkov, V. Y., Lorius, C., PÉpin, L., Ritz, C., Saltzman, E. and Stievenard, M. 1999. Climate and atmospheric history of the past 420,000 years from the Vostok ice core, Antarctica. – Nature 399: 429-436. Petit, R. J. and Hampe, A. 2006. Some evolutionary consequences of being a tree. – Annual Review of Ecology, Evolution, and Systematics 37: 187-214. Pickard, W. F. 1983. Three interpretations of the self-thinning rule. – Annals of Botany 51: 749- 757. Pickett, S. T. A. 1980. Non-equilibrium coexistence of plants. – Bulletin of the Torrey Botanical Club 107: 238. Pickett, S. T. A., Collins, S. L. and Armesto, J. J. 1987. Models, mechanisms and pathways of succession. – The Botanical Review 53: 335-371. Prentice, C. I., Bondeau, A., Cramer, W., Harrison, S. P., Hickler, T., Lucht, W., Sitch, S., Smith, B. and Sykes, M. 2007. Dynamic global vegetation modeling: Quantifying terrestrial ecosystem responses to large-scale environmental change. – In: Canadell, J. G. et al. (eds.), Terrestrial ecosystems in a changing world. Springer, pp. 175–192. Pretzsch, H., Biber, P. and Ďurský, J. 2002. The single tree-based stand simulator SILVA: Construction, application and evaluation. – Forest Ecology and Management 162: 3-21. Pretzsch, H., Bielak, K., Block, J., Bruchwald, A., Dieler, J., Ehrhart, H.-P., Kohnle, U., Nagel, J., Spellmann, H., Zasada, M. and Zingg, A. 2013. Productivity of mixed versus pure stands of oak (Quercus petraea (Matt.) Liebl. and Quercus robur L.) and European beech (Fagus sylvatica L.) along an ecological gradient. – European Journal of Forest Research. 28
References Pretzsch, H., Dieler, J., Matyssek, R. and Wipfler, P. 2010. Tree and stand growth of mature Norway spruce and European beech under long-term ozone fumigation. – Environmental Pollution 158: 1061-1070. Pretzsch, H., Grote, R., Reineking, B., Rötzer, T. and Seifert, S. 2008. Models for forest ecosystem management: A European perspective. – Annals of Botany 101: 1065-1087. Purves, D. W. and Pacala, S. 2008. Predictive models of forest dynamics. – Science 320: 1452- 1453. Rastetter, E. B., King, A. W., Cosby, B. J., Hornberger, G. M., O'Neill, R. V. and Hobbie, J. E. 1992. Aggregating fine-scale ecological knowledge to model coarser-scale attributes of ecosystems. – Ecological Applications 2: 55. Rehfeldt, G. E., Ferguson, D. E. and Crookston, N. L. 2009. Aspen, climate, and sudden decline in western USA. – Forest Ecology and Management 258: 2353-2364. Rivas-Martínez, S., Penas, A. and Díaz, T. E. 2004. Biogeographic Map of Europe – Cartographic Service. Rodríguez-Sánchez, F., Hampe, A., Jordano, P. and Arroyo, J. 2010. Past tree range dynamics in the Iberian Peninsula inferred through phylogeography and palaeodistribution modelling: A review. – Review of Palaeobotany and Palynology 162: 507-521. Rossi, S., Deslauriers, A., Anfodillo, T., Morin, H., Saracino, A., Motta, R. and Borghetti, M. 2006. Conifers in cold environments synchronize maximum growth rate of tree-ring formation with day length. – New Phytologist 170: 301-310. Ruiz-Benito, P., Gómez-Aparicio, L. and Zavala, M. A. 2012. Large-scale assessment of regeneration and diversity in Mediterranean planted pine forests along ecological gradients. – Diversity and Distributions 18: 1092-1106. Ruiz-Benito, P., Lines, E. R., Gómez-Aparicio, L., Zavala, M. A., Coomes, D. A. and Hector, A. 2013. Patterns and drivers of tree mortality in Iberian forests: Climatic effects are modified by competition. – PLoS ONE 8: e56843. Sala, A., Piper, F. and Hoch, G. 2010. Physiological mechanisms of drought-induced tree mortality are far from being resolved. – New Phytologist 186: 274-281. Salmond, J. and Smith, C. G. 1996. Back to basics: World climatic types. – Weather 51: 11-18. Sánchez-Gómez, D., Zavala, M. A., Schalkwijk, D. B., Urbieta, I. R. and Valladares, F. 2008. Rank reversals in tree growth along tree size, competition and climatic gradients for four forest canopy dominant species in central Spain. – Annals of Forest Science 65: 605. Scharnweber, T., Manthey, M., Criegee, C., Bauwe, A., Schröder, C. and Wilmking, M. 2011. Drought matters – Declining precipitation influences growth of Fagus sylvatica L. and Quercus robur L. in north-eastern Germany. – Forest Ecology and Management 262: 947-961. Scholze, M. 2006. A climate-change risk analysis for world ecosystems. – Proceedings of the National Academy of Sciences 103: 13116-13120. Schroter, D. 2005. Ecosystem service supply and vulnerability to global change in Europe. – Science 310: 1333-1337. Schumacher, S., Bugmann, H. and Mladenoff, D. 2004. Improving the formulation of tree growth and succession in a spatially explicit landscape model. – Ecological Modelling 180: 175 194.‐ Schumacher, S., Reineking, B., Sibold, J. and Bugmann, H. 2006. Modeling the impact of climate and vegetation on fire regimes in mountain landscapes. – Landscape Ecology 21: 539-554. Sexton, J. P., McIntyre, P. J., Angert, A. L. and Rice, K. J. 2009. Evolution and ecology of species range limits. – Annual Review of Ecology, Evolution, and Systematics 40: 415-436. 29
References Shugart, H. H. and West, D. C. 1981. Long-term dynamics of forest ecosystems. – American Scientist 69: 647-652. Sitch, S., Huntingford, C., Gedney, N., Levy, P. E., Lomas, M., Piao, S. L., Betts, R., Ciais, P., Cox, P., Friedlingstein, P., Jones, C. D., Prentice, C. I. and Woodward, F. I. 2008. Evaluation of the terrestrial carbon cycle, future plant geography and climate-carbon cycle feedbacks using five Dynamic Global Vegetation Models (DGVMs). – Global Change Biology 14: 2015-2039. Soja, A. J., Tchebakova, N. M., French, N. H. F., Flannigan, M. D., Shugart, H. H., Stocks, B. J., Sukhinin, A. I., Parfenova, E. I., Chapin, F. S. and Stackhouse, J. P. W. 2007. Climate-induced boreal forest change: Predictions versus current observations. – Global and Planetary Change 56: 274-296. Sousa, W. 1984. The role of disturbance in natural communities. – Annual Review of Ecology and Systematics 15: 353-391. Starfield, A. M. 1997. A pragmatic approach to modeling for wildlife management. – The Journal of Wildlife Management 61: 261. Street-Perrott, A. et al. (eds.) 1983. Variations in the global water budget – Springer Netherlands. Stueve, K. M., Perry, C. H., Nelson, M. D., Healey, S. P., Hill, A. D., Moisen, G. G., Cohen, W. B., Gormanson, D. D. and Huang, C. 2011. Ecological importance of intermediate windstorms rivals large, infrequent disturbances in the northern Great Lakes. – Ecosphere 2: art2. Sunderlin, W. D., Dewi, S., Puntodewo, A., Mueller, D., Angelsen, A. and Epprecht, M. 2008. Why forests are important for global poverty alleviation: A spatial explanation. – Ecology and Society 13. Svenning, J.-C. and Skov, F. 2007. Could the tree diversity pattern in Europe be generated by postglacial dispersal limitation? – Ecology Letters 10: 453-460. Sykes, M. and Prentice, C. I. 1996. Climate change, tree species distributions and forest dynamics: A case study in the mixed conifer/northern hardwoods zone of northern Europe. – Climatic Change 34. Taberlet, P., Fumagalli, L., Wust-Saucy, A.-G. and Cosson, J.-F. 1998. Comparative phylogeography and postglacial colonization routes in Europe. – Molecular Ecology 7: 453-464. Temperli, C., Bugmann, H. and Elkin, C. 2012. Adaptive management for competing forest goods and services under climate change. – Ecological Applications 22: 2065-2077. Theurillat, J.-P. and Guisan, A. 2001. Potential impact of climate change on vegetation in the European Alps: A review. – Climatic Change 50: 77-109. Thiel, D., Nagy, L., Beierkuhnlein, C., Huber, G., Jentsch, A., Konnert, M. and Kreyling, J. 2012. Uniform drought and warming responses in Pinus nigra provenances despite specific overall performances. – Forest Ecology and Management 270: 200-208. Thuiller, W., Albert, C., Araújo, M. B., Berry, P. M., Cabeza, M., Guisan, A., Hickler, T., Midgley, G. F., Paterson, J., Schurr, F. M., Sykes, M. and Zimmermann, N. E. 2008. Predicting global change impacts on plant species’ distributions: Future challenges. – Perspectives in Plant Ecology, Evolution and Systematics 9: 137-152. Thuiller, W., Vayreda, J., Pino, J., Sabate, S., Lavorel, S. and Gracia, C. 2003. Large-scale environmental correlates of forest tree distributions in Catalonia (NE Spain). – Global Ecology and Biogeography 12: 313-325. Tilman, D. 1987. The importance of the mechanisms of interspecific competition. – The American Naturalist 129: 769. Tilman, D. 1994. Competition and biodiversity in spatially structured habitats. – Ecology 75: 2. 30
References Trenberth, K. E. 1999. Atmospheric moisture recycling: Role of advection and local evaporation. – Journal of Climate 12: 1368-1381. Turchin, P. 2001. Does population ecology have general laws? – Oikos 94: 17-26. Turner, M. G., Baker, W. L., Peterson, C. J. and Peet, R. K. 1998. Factors influencing succession: Lessons from large, infrequent natural disturbances. – Ecosystems 1: 511-523. Turner, R. K. and Daily, G. C. 2008. The ecosystem services framework and natural capital conservation. – Environmental and Resource Economics 39: 25-35. Underwood, A. J. 2009. Components of design in ecological field experiments. – Annales Zoologici Fennici 46: 93-111. Urbieta, I. R., Pérez-Ramos, I. M., Zavala, M. A., Marañón, T. and Kobe, R. K. 2008. Soil water content and emergence time control seedling establishment in three co-occurring Mediterranean oak species. – Canadian Journal of Forest Research 38: 2382-2393. van der Maarel, E., Noest, V. and Palmer, M. W. 1995. Variation in species richness on small grassland quadrats: Niche structure or small-scale plant mobility? – Journal of Vegetation Science 6: 741-752. van Oijen, M., Reyer, C., Bohn, F., Cameron, D., Deckmyn, G., Flechsig, M., Härkönen, S., Hartig, F., Huth, A., Kiviste, A., Lasch, P., Mäkelä, A., Mette, T., Minunno, F. and Rammer, W. 2013. Bayesian calibration, comparison and averaging of six forest models, using data from Scots pine stands across Europe. – Forest Ecology and Management 289: 255-268. van Oijen, M., Rougier, J. and Smith, R. 2005. Bayesian calibration of process-based forest models: bridging the gap between models and data. – Tree Physiology 25: 915-927. Veblen, T. T., Hadley, K. S., Nel, E. M., Kitzberger, T., Reid, M. and Villalba, R. 1994. Disturbance regime and disturbance interactions in a Rocky Mountain subalpine forest. – Journal of Ecology 82: 125-135. Veloz, S. D., Williams, J. W., Blois, J. L., He, F., Otto-Bliesner, B. and Liu, Z. 2012. No-analog climates and shifting realized niches during the late quaternary: Implications for 21st-century predictions by species distribution models. – Global Change Biology 18: 1698-1713. Wardle, P. 1964. Facets of the distribution of forest vegetation in New Zealand. – New Zealand Journal of Botany 2: 352 366.‐ Webster, M., Forest, C., Reilly, J., Babiker, M., Kicklighter, D., Mayer, M., Prinn, R., Sarofim, M., Sokolov, A., Stone, P. and Wang, C. 2003. Uncertainty analysis of climate change and policy response. – Climatic Change 61: 295-320. Weiner, J. and Thomas, S. C. 2001. The nature of tree growth and the ‘age-related decline in forest productivity’. – Oikos 94: 374-376. Weisberg, P. J. and Bugmann, H. 2003. Forest dynamics and ungulate herbivory: From leaf to landscape. – Forest Ecology and Management 181: 1-12. Wermelinger, B. 2004. Ecology and management of the spruce bark beetle Ips typographus - A review of recent research. – Forest Ecology and Management 202: 67-82. White, P. S. and Jentsch, A. 2001. The search for generality in studies of disturbance and ecosystem dynamics. – Progress in Botany 63: 399-449. White, P. S. and Pickett, S. T. A. 1985. Natural disturbance and patch dynamics: An introduction. – In: Pickett, S. T. A. (ed.), The ecology of natural disturbance and patch dynamics. Academic Pr., pp. 3–13. Whitmore, T. C. 1989. Canopy gaps and the two major groups of forest trees. – Ecology 70: 536. 31
References Williams, J. W. and Jackson, S. T. 2007. Novel climates, no-analog communities, and ecological surprises. – Frontiers in Ecology and the Environment 5: 475-482. Winter, S., Chirici, G., McRoberts, R. E., Hauk, E. and Tomppo, E. 2008. Possibilities for harmonizing national forest inventory data for use in forest biodiversity assessments. – Forestry 81: 33-44. Wissel, C. 1989. Goals and possibilities of theoretical ecology, as illustrated by the island theory. – Verhandlungen Gesellschaft für Oekologie 18: 483-490. Wissel, C. 1992. Aims and limits of ecological modelling exemplified by island theory. – Ecological Modelling 63: 1-12. Worrall, J. J., Egeland, L., Eager, T., Mask, R. A., Johnson, E. W., Kemp, P. A. and Shepperd, W. D. 2008. Rapid mortality of Populus tremuloides in southwestern Colorado, USA. – Forest Ecology and Management 255: 686-696. Wortemann, R., Herbette, S., Barigah, T. S., Fumanal, B., Alia, R., Ducousso, A., Gomory, D., Roeckel-Drevet, P. and Cochard, H. 2011. Genotypic variability and phenotypic plasticity of cavitation resistance in Fagus sylvatica L. across Europe. – Tree Physiology 31: 1175-1182. Wunder, J., Bigler, C., Reineking, B., Fahse, L. and Bugmann, H. 2006. Optimisation of tree mortality models based on growth patterns. – Ecological Modelling 197: 196-206. Wunder, J., Brzeziecki, B., Zybura, H., Reineking, B., Bigler, C. and Bugmann, H. 2008. Growthmortality relationships as indicators of life-history strategies: a comparison of nine tree species in unmanaged European forests. – Oikos 117: 815-828. Xu, C., Gertner, G. Z. and Scheller, R. M. 2009. Uncertainties in the response of a forest landscape to global climatic change. – Global Change Biology 15: 116-131. Yamamoto, S.-I. 2000. Forest gap dynamics and tree regeneration. – Journal of Forest Research 5: 223-229. Yao, J., He, X., Wang, A., Chen, W., Li, X., Lewis, B. J., Lv, X. and Bond-Lamberty, B. 2012. Influence of forest management regimes on forest dynamics in the upstream region of the Hun river in northeastern China. – PLoS ONE 7: e39058. Zang, C., Pretzsch, H. and Rothe, A. 2012. Size-dependent responses to summer drought in Scots pine, Norway spruce and common oak. – Trees 26: 557-569. Zeide, B. 1993. Analysis of growth equations – Forest Science 39: 594-616. 32
Manuscripts Manuscripts Summary of the following manuscripts The first study was realized in a mountain forest of New Zealand along an altitudinal gradient. In Manuscript 1 ‘Long-term vegetation dynamics in New Zealand’, forest succession after a catastrophic volcanic eruption at the slope of Mt. Hauhungatahi, North Island, was simulated using the forest simulation model LandClim. The study explored climate and competition as drivers of observed forest dynamics and calibrated LandClim for the first time for New Zealand´s temperate mountain forest. Manuscript 2 ‘Climatic turning point for beech and oak’ directly addressed the of question whether contemporary forests will change during the next decades. The study region was located in south-western Germany in a climatically dry region. Employing two different forest simulation models (SILVA and LandClim), the response of a mixed beech-oak stand was assessed under non-stationary climate. The results showed that a ‘climatic turning point’ in terms of species dominance might be passed during the present century. Inherited stand structure, however, might lead to a considerable time delay of observable forest response in undisturbed stands. The combined effect of climate and competition on tree growth was investigated in more detail in a study based on the Spanish National Inventories. In Manuscript 3 ‘Symmetric and asymmetric competition’, changes of competition intensity and importance along the aridity gradient throughout the Iberian Peninsula were assessed. Symmetric and asymmetric competition were distinguished and integrated within a statistical model. The study exhibited that the negative effect of competition on tree growth increased with aridity. Manuscript 4 ‘Ecotype mixing as climate change adaptation’ deals with the question how to adapt forests to climate change. One management strategy might be to introduce non-local ecotypes in order to increase biodiversity. Thereby, future yield is assumed to be more stable, according to the portfolio effect theory. However, forest stands differ from common economic products by the natural process of self-thinning. A consideration of self-thinning in the evaluation of ecotype 33
Manuscript 1: Long-term vegetation dynamics in New Zealand a function of limiting environmental factors and biotic interactions. Despite the problem of determining the maximum potential growth rate (parameter ‘Rmax’, see Tab. 1), this down-weighting approach is found in most forest gap models, such as JABOWA (Bugmann 2001). Tree mortality is a function of three factors: Growth-dependent stress, density-dependent stress and an intrinsic, age-related component. Detailed descriptions of the formulation of stand-scale processes in LandClim are given in Supplement S1. LandClim represents the effects of disturbance on forest composition and structure reciprocally, which, in turn, allows for studies of past and future changing environments (Schumacher and Bugmann 2006, Henne et al. 2011). Environmental input variables are topography, soil water capacity and ‘land-type’ (a userdefined map assigning specific establishment rates and disturbance regimes to distinct areas), as well as temperature and precipitation at monthly resolution. Spe- 40 Tab. 1: Species life-history parameters. Tolerance-classes range from 1 (lowest) to 5 (highest tolerance). Abbreviations: EG: Evergreen, BL-EG: Broadleaf-evergreen. A brief explanation as well as references and basis for the parameter choice are given in Supplement, S2 and S3. The parameters ‘shade tolerance’ and ‘Rmax’ were determined via a pattern-oriented parameterization. The parameter ‘minDD’ was calibrated to fit species upper altitudinal limits described in Druitt et al. (1990), see also Supplement S2 for further details. Drought-, fire- and browsing-tolerance were not relevant in the present study, therefore, a default value of 3 was assigned to all species. Parameters are discussed in more detail in Schumacher (2004). Parameter name Parameter description Dacrydium cupressinum Libocedrus bidwillii Podocarpus unninghamii Tree fern Weinmannia racemosa maxAge Maximum age (years) that an individual can reach 800 1000 650 150 400 Kmax Maximum aboveground tree biomass (tons) a species can reach 12 8 7.73 0.5 6.32 leafHabit Leaf habit (form) EG EG EG BL-EG BL-EG foliageType Shading potential of a species canopy 33455 minTemperature Minimum temperature (°C) for establishment -8 -13 -13 -8 -8 shadeTolerance Species shade tolerance 4 2 2 5 4 droughtTolerance Species drought tolerance 3 3 3 3 3 minDD Minimum annual degree day sum 1400 1200 1280 1550 1300 Rmax Maximum above-ground biomass growth rate (per year) 0.07 0.12 0.11 0.10 0.13
Manuscript 1: Long-term vegetation dynamics in New Zealand cies environmental responses are defined by traits such as drought tolerance, shade tolerance and temperature requirements. LandClim operates over long time scales (hundreds to thousands of years) and large spatial extents (100s to 10000s of hectares) at a relatively fine scale (grid cells of 25 by 25 m). It has previously been tested in and adapted to the European Alps, the North American Rocky Mountains and Mediterranean forests (Schumacher and Bugmann 2006, Colombaroli et al. 2010, Henne et al. 2011, Henne et al. 2012, Briner et al. 2012, Elkin et al. 2012). This study is the first application of LandClim to the Southern Hemisphere. The model structure of LandClim was kept unchanged; the only differences from the previous studies were a new allometric relationship for calculation of NZ’s species biomass to DBH, an increase in the maximum stand biomass from 300 t/ha to 1000 t/ha (see Supplement S1 for details) and the parameterization of tree species. Study site Our simulations focused on the western slope of Mt. Hauhungatahi, which has been intensively studied previously (Druitt et al. 1990, Horrocks and Ogden 1998, Ogden et al. 2005). Mt. Hauhungatahi is located in the Tongariro National Park in the central North Island of New Zealand (Fig. 1) where forests have been subject to recurrent volcanic events throughout the Quaternary, with therhyolitic Taupo eruption of 1718 ± 5 cal. BP (Hogg et al. 2012) particularly significant (Horrocks and Ogden 1998). Druitt et al. (1990) distinguished three main belts of forest 41 Fig. 1: Map of study area on the western slope of Mt. Hauhungatahi in the Tongariro National Park, New Zealand.
Manuscript 1: Long-term vegetation dynamics in New Zealand using the importance value as a measure of species dominance (species nomenclature follows Landcare Research 2011b): (i) A montane forest from 850 to 1000 m a.s.l. with the canopy dominated by angiosperms, in particular Weinmannia racemosa (kamāhi). Scattered old conifer individuals (e.g. Dacrydium cupressinum – rimu) are present and constitute an important part of the total basal area, but younger conifer individuals are mostly absent. Tree ferns (e.g. Cyathea smithii – katote) are important components of the sub-canopy layer. (ii) A transitional zone ranging from 1000 to 1050 m a.s.l., where several species (including W. racemosa, D. cupressinum and tree ferns) reach their upper limit. Conifers, and in particular Podocarpus cunninghamii (formerly P. hallii, Hall’s totara), are prominent in this belt. (iii) A sub-alpine zone (from 1050 m to the treeline), which is largely dominated by the conifer Libocedrus bidwillii (pāhautea). The treeline (formed by L. bidwillii, together with Halocarpus biformis) is highly discontinuous, varying between altitudinal limits of 1100 to 1250 m a.s.l. Ogden et al. (2005) reported the highest densities and most vigorous regeneration of L. bidwillii in the sub-alpine zone, with densities and regeneration declining towards the transition zone at around 1050 m. This decline coincided with an increase in angiosperm densities, with W. racemosa becoming dominant in the upper montane zone. Horrocks and Ogden (1998) outline the effect of the Taupo eruption on forest succession at Mt. Hauhungatahi using detailed pollen diagrams collected along the altitudinal transect described by Druitt et al. (1990). The two dominant patterns seen in Horrocks and Ogden’s pollen diagrams are: (i) the initial increase and spread of L. bidwillii immediately following the eruption, and (ii) the progressive spread of the angiosperm W. racemosa in the montane forest in the centuries following the eruption, resulting in the present vegetation zonation. Simulation experiments The western slope of Mt. Hauhungatahi was represented in LandClim on a grid of cells (25 × 25 m) describing topography and soils. Topographic parameters (elevation, aspect and slope) were derived from a digital elevation map of the area 42
Manuscript 1: Long-term vegetation dynamics in New Zealand (DEM25, Land Information New Zealand 2011). Areas below 780 m were excluded since they are currently deforested and could not, therefore, be compared with empirical data. The widespread peatland area above the treeline (> 1300 m) was also excluded. Soil characteristics on the slopes of Mt. Hauhungatahi are spatially heterogeneous and are a function of volcanic activity and a suite of secondary processes (Druitt et al. 1990). Due to the generally high water-holding capacity of volcanic soils (Scheffer and Schachtschabel 2002), a high soil water capacity (bucket size of 200 mm) was assigned uniformly to all grid cells (the model was not sensitive to this assumption, results not shown). The climate record (1930-2000 AD) from the nearby Chateau climate station at Mt. Ruapehu (12 km east of Mt. Hauhungatahi, 1097 m a.s.l.) was used as the climatic input data for the model (data source: NIWA 2011). Because the temperature reconstructions available for the site (Palmer and Xiong 2004) do not span the full 1700 year succession that we consider, the simulation was carried out under a present climate to provide a parsimonious baseline scenario. While climate reconstructions suggest the existence of some warm- and cold-periods in the past centuries, these are only of the order of less than 0.5 °C (Palmer and Xiong 2004). Since climate change is generally considered to be small in the post-Taupoperiod compared to the millennium before the eruption (McGlone 1989, Rogers and McGlone 1989) and the fossil pollen data do not show evidence of climate-related vegetation change after the eruption (Horrocks and Ogden 1998), we considered our assumption as reasonably realistic. The 70 year climate record was resampled (with replacement) to generate a 1700 year climate sequence; this randomisation was repeated for every simulation. The four dominant canopy species that characterise the altitudinal belts on Mt Hauhungatahi were represented in the model: The angiosperm Weinmannia racemosa (kamāhi) and the conifers Dacrydium cupressinum (rimu), Podocarpus cunninghamii (Hall’s totara) and Libocedrus bidwillii (pāhautea). A tree fern lifeform was included due to their high abundance in the montane forest (especially Cyathea smithii ;Druitt et al. 1990) and their structural importance for forest dynamics (Coomes et al. 2005). The tree-fern life-form was implemented as a shade tolerant understory species (resembling the behaviour of Cyathea smithii as reported by Bystriakova et al. 2011) and treated by the model in the same way as the other tree species. 43
Manuscript 1: Long-term vegetation dynamics in New Zealand Species life-history traits were assigned from the New Zealand Eco-Trait database (Landcare Research 2011a), the ecological literature, including the Flora of New Zealand, and expert-knowledge. Details about the life-history traits and reasoning for the choice of parameters are provided in Supplement S2. The parameter ‘minDD’ (minimum degree days) was calibrated to fit the upper altitudinal limits described in Druitt et al. (1990), assuming that the species upper elevation limit is controlled by temperature (see Supplement S2 for further information). The parameters ‘shade tolerance’ and ‘Rmax’ (maximum relative biomass growth rate per year) were determined in a pattern-oriented parametrization approach, described in the following section. It was assumed that the Taupo eruption removed all vegetation from the study area, since the actual degree of forest destruction remains unknown. Horrocks and Ogden (1998) noted that the effect of the Taupo eruption (including the shock wave, air fall of tephra and subsequent fires) was presumably considerable at Mt. Hauhungatahi (ca. 75 km distance from the vent), although fossil pollen indicate some surviving forest in the area. The simulations started from ‘bare ground’, initiated by a spatially homogeneous seed rain with the same amount of seed across all species, as is often assumed in forest landscape and gap models, particularly in the absence of detailed species-specific empirical data describing the seed rain. Although this approach neglects the possible influences of prior vegetation and heterogeneous seed rain, it provides a baseline assumption to cope with complex and unknown initial conditions. For the same reason, global seed dispersal was assumed. Succession was simulated and tracked over 1700 years, representing the time since the Taupo eruption. No further disturbance events were simulated over this succession. To ensure direct comparability with the measurements of forest structure by Ogden et al. (2005), only individuals with a DBH (diameter at breast height) larger than 10 cm were considered in the analyses of the simulated altitudinal transect (although LandClim is capable of tracking smaller individuals). Pattern-oriented parameterization In order to identify plausible parameter combinations for ‘maximum growthrate’(Rmax) and ‘shade tolerance’ that reproduce the expected system behaviour, a pattern-oriented parameterization approach was chosen. 44
Manuscript 1: Long-term vegetation dynamics in New Zealand First, a complete parameter space with both traits for all five species was created. Each parameter was given five possible values (for shade tolerance from 1 (low) to 5 (high) and ‘Rmax’ from 0.03 to 0.15 (in steps of 0.03)). This parameter space was reduced by discarding ecologically unreasonable parameter combinations a priori from the analysis, using the criteria below: (i) Maximum growth rates of angiosperms (W. racemosa) should be higher than those of conifers (Ogden and Stewart 1995). (ii) Shade tolerance of conifers L. bidwillii and P. cunninghamii should be in the range from low to intermediate due to their characterisation as light-demanding pioneer species (Clayton-Greene 1977, Ebbett and Ogden 1998). (iii) Shade tolerance of D. cupressinum and W. racemosa should range between intermediate to high (Lusk and Ogden 1992, Lusk et al. 2009). (iv) Shade tolerance of the tree fern life form (resembling the ecology of Cyathea smithii) was defined as high (see Supplement S3 for further details about the tree fern life form). The model was run for all possible parameter combinations that fulfilled these criteria (in total 23364 scenarios). The simulations for the pattern-oriented parameterization were performed on a small area (covering the length of the entire altitudinal gradient but reducing the width to only 4 cells, i.e. 100 m) in order to reduce computation time for each simulation, under a randomized present climate scenario in the absence of further disturbance events. The simulation outputs were filtered according to spatio-temporal patterns described in the pollen records of Horrocks and Ogden (1998), using a Boolean filter. These patterns were (i) an initial dominance of conifers (in particular L. bidwillii) in the first centuries after the Taupo eruption which was defined as the criterion that conifer species should reach > 60% of total forest biomass in the first 300 years of succession; (ii) a progressive spread of W. racemosa in the montane area during later stages of succession, represented by the criterion that W. racemosa should reach > 60% of total forest biomass during the years 1000-1700 in the area up to 1050 m altitude. Additionally, parameter combinations resulting in unreasonably low total stand biomass were discarded. Based on the carbon stock estimates for podocarp-hardwood forests in NZ (Horrocks and Ogden 1998), a minimum biomass of 100 t per ha was estimated (under the assumption that biomass consist of 50% carbon). Further parameter combinations for which tree species had disappeared at the end of succession (i.e. species biomass 45
Manuscript 1: Long-term vegetation dynamics in New Zealand dropped below < 1% of total biomass) were also discarded. These criteria narrowed down the parameter space substantially. Finally, a refined sensitivity analysis was performed to determine optimized ‘Rmax’ values for each species. ‘Rmax’ values were assigned from within the range determined by the previous analyses and sampled in steps of 0.01. Shade tolerance values were assigned according to the results of the previous analysis. For the filtering of these results, stronger criteria were applied: L. bidwillii should dominate the early successional stage with > 75% of total standing biomass, and W. racemosa should dominate the late successional stage in the montane forest with > 75% biomass. Furthermore, simulated mean annual DBH-growth of species (at year 1700) should be in the range of mean annual growth rates reported by Ogden et al. (2005) for L. bidwillii, Smale and Smale (2003) for P. cunninghamii, Lusk and Ogden (1992) for D. cupressinum and W. racemosa (see Supplement S2). For the tree fern life form, height-growth estimates from Ogden et al. (1997a) were used (see Supplement S3). The final parameter set was used to simulate the spatio-temporal forest dynamics of Mt. Hauhungatahi (Tab. 1). Simulations were repeated 50 times with a randomized present climate to account for stochastic variation between model realisations under the same parameter conditions. Analyses and visualization of model results were conducted using R.2.15.2 (R Development Core Team 2012). The pattern-oriented parameterization was carried out on a high-performance computer cluster at the University of Bayreuth, Germany. Results Species traits In the pattern-oriented parameterization, the application of the filter criteria (see previous section) narrowed down the parameter space from 23364 to 129 possible combinations. The shade tolerance ranges defined a priori proved suitable, as most model results from the given range were accepted by the filter criteria (Fig. 2, black bars). The specific ‘Rmax’ values substantially influenced the successional sequence and the biomass of individual species. ‘Rmax’ values below 0.06 typically resulted in low species biomass and therefore tended to be 46
Manuscript 1: Long-term vegetation dynamics in New Zealand discarded by the filter criteria. The refined sensitivity analysis showed that the observed successional dynamic was reproduced within a rather narrow constellation of ‘Rmax’ values. L. bidwillii and W. racemosa (as the main components of the simulated forest ecosystem) displayed a strong successional differentiation between pioneer and later successional species once the parameter constellation was set as shown in Fig. 2. With increasing advantage in growth rate of W. racemosa over L. bidwillii, the pattern still prevailed but the initial dominance of L. bidwillii became less pronounced at the expense of W. racemosa in the lower montane forest. Within the parameter ranges explored, P. cunninghamii and D. cupressinum were present with only low biomass. Variations in their parameter sets consequently had a minimal effect on the gross successional trends. Species zonation LandClim reproduced the actual species zonation described by Druitt et al. (1990) at the end of succession (year 1700) once all species (with traits as in Tab. 1) were included in competition with each other. The species upper elevation limits were controlled by their specific temperature (‘minimum degree day’) requirements. Notably, the observations of Druitt et al. (1990) were used in the calibration of the ‘minimum degree day’ parameter, therefore they cannot be con- 47 Fig. 2: Results of the pattern-oriented parameterization for the parameters ‘shade tolerance’ (left) and maximum growth rate - ‘Rmax’ (right). The range of accepted parameters is indicated by the black line, parameter values occurring with highest frequencies are indicated by a circle (filled circle indicates result of refined sensitivity analysis of ‘Rmax’).
Manuscript 1: Long-term vegetation dynamics in New Zealand sidered as independent data for model evaluation. In simulations of monocultures each species occurred across the altitudinal transect from the lower boundary (780 m a.s.l.) to its specific upper elevation limit. Once all species were included (and thus interspecific competition occurred), conifer species largely disappeared from lower altitudes. The lower elevation forest band (780-950 m) was dominated by W. racemosa. Tree ferns occurred up to an altitude of around 880-900 m, but with a low basal area (< 1m² / ha) and density (< 50 stems/ha, data not shown). D. cupressinum occurred up to an altitude of 1000 m, but only as a very few, scattered individuals. Above 950 m a.s.l., the basal area of W. racemosa steadily declined to its upper altitudinal limit at around 1050 m. As W. racemosa declined, the forest gradually shifted in composition with the conifer species becoming increasingly important. P. cunninghamii reached its maximum basal area and stem density between altitudes of 950 to 1050 m. Altitudes above 1000 m were dominated by L. bidwillii in terms of basal area and stem density until the treeline at around 1220 m. The simulated basal area and stem density of the species peaked at the same altitudes as those described by Ogden et al. (2005), but the model under-repres- ented basal area for all species and over represented stem density for L. bidwillii (Fig. 3). The deviation between observed and simulated forest structure was particularly evident for D. cupressinum and P. cunninghamii, which achieved only very low values of basal area and stem density in the simulated year 1700. Forest succession The simulated post-Taupo succession resembled the general patterns described by Horrocks and Ogden (1998) once species life-history traits were assigned following a rigorous and thorough pattern-oriented parameterization process (Tab. 1). The succession was characterised by three main stages, one from simulation years 0-400, a second from years 400-700, and a third from year 700 onwards. During the early phase of the succession (simulation years 0 to 400), L. bidwillii dominated the study area in terms of biomass (Fig. 4). In the following centuries (simulation years 400-700, Fig. 4), the biomass of L. bidwillii declined, whereas that of W. racemosa steadily increased. From around simulation year 1000 onwards, conditions were generally stable and similar to the contemporary composition on Mt. Hauhungatahi. Both D. cupressinum and P. cunning- 48
Manuscript 1: Long-term vegetation dynamics in New Zealand hamii were only present with low biomass throughout the succession. The tree fern life-form occurred across the entire succession, but at lower biomasses than the other four tree species. The spatial distribution of the tree species shifted over time such that there was a progressive upward expansion of W. racemosa in the montane forest, accompanied by a retraction of L. bidwillii and P. cunninghamii into the higher sub-alpine forest. The early successional stage (simulation years 0 to 400) was characterised by the widespread dominance of L. bidwillii (Fig. 4). In the montane forest, the dominance of L. bidwillii was associated with occasional occurrence of P. cunninghamii and D. cupressinum. W. racemosa occurred only in a few scattered grid cells during the early succession. During the mid-succession (simu- 49 Fig. 3: Altitudinal distribution of species as described by Druitt et al. (1990) (shown as grey bar) and forest structure (basal area and density) as reported by Ogden et al. (2005) (shown as observation points) alongside model outcomes after 1700 simulation years. For model outcomes, the mean of 50 model realisations is shown together with the standard error of the mean. Only individuals with DBH > 10 cm were considered in the model in order to assure comparability with the empirical data. On the basis of their low abundance and the lack of empirical data reporting their stand structure, tree ferns were excluded.
Manuscript 1: Long-term vegetation dynamics in New Zealand c. 3000-2000 BP (McGlone and Moar 1977, Rogers and McGlone 1989) might have meant that L. bidwillii was expanding at the time of the eruption. According to Horrocks and Ogden (1998), the effects of the Taupo eruption on Mt. Hauhungatahi were patchy, with some areas of forest escaping damage. Surviving patches of forest could have had a substantial impact on the vegetation composition immediately after the eruption, similar to contemporary post-disturbance succession at Mt. St. Helens (Dale 2005). Despite its simplified assumptions (succession starting from bare ground with global and uniform seed dispersal), the model results agree with the key patterns seen in the palynological record. The model analysis therefore highlights the profound and sustained effect of a catastrophic disturbance event such as the Taupo eruption for long-lived pioneer conifer species such as L. bidwillii (see also Ogden et al. 2005). More differentiated patterns in the altitudinal series of pollen assemblages are, however, difficult to compare to the model results. First, the model represented only the main canopy species and did not account for other understorey species, some of which make substantial contributions to the relative abundance of pollen. Second, differences in pollen preservation and dispersal between species mean that there is not a consistent relationship between modern pollen abundance and basal area for all simulated species at Mt. Hauhungatahi (Ogden et al. 2005), which, in turn, makes it challenging to directly relate pollen assemblages to the simulated forest structure. Application of LandClim for the NZ context LandClim’s facilities to represent disturbance processes (e.g. fires and standreplacing windstorms) make it particularly well-suited for exploring questions about vegetation dynamics across broad-scales in space and time. This, in turn, means that LandClim can help to address the long-standing questions surrounding the role of disturbance processes in angiosperm-gymnosperm coexistence in NZ’s forests (as suggested by a number of empirical studies, e.g. Lusk and Ogden 1992, Wells et al. 2001, Ogden et al. 2005). On the basis of a pattern-oriented parameterization of species traits and no other structural changes, LandClim proved capable of reproducing the altitudinal distribution of species on Mt. Hauhungatahi described by Druitt et al. (1990) and, to some extent, the patterns observed in forest structure by Ogden et al. (2005). Furthermore, the model experiments demonstrate how tree species life-history traits might explain the patterns of suc- 56
Manuscript 1: Long-term vegetation dynamics in New Zealand cession seen in the palynological record (Horrocks and Ogden 1998). On the other hand, relative to the data of Ogden et al. (2005), LandClim systematically underestimates basal area and tends to overestimate stem density for some species (Fig. 3). This mismatch suggests that the regeneration and mortality of long-lived trees is not represented adequately in LandClim. A more thorough consideration of how regeneration is represented in the model will be an important component of LandClim’s development for future application in New Zealand. Tree ferns constitute a distinctive feature of NZ’s forests that have no direct equivalent in European and North American forest ecosystems. Our study provides a first attempt to incorporate these into a forest landscape model, but a more adequate representation will need to account for their distinctive growth behaviour. Conclusion Our approach highlights the potential for combining forest-landscape modelling with palaeoecological reconstructions in spatially complex environments. The use of simulation models to explore drivers underlying long-term dynamics observed in palaeoecological reconstructions is an area of considerable current interest. Whereas previous such studies using LandClim (Henne et al. 2011) have focused on cumulative pollen abundances over entire catchments, our study shows the model’s suitability for use with locally and regionally distinct pollen assemblages. In the forest landscapes we consider, vegetation dynamics are controlled by interactions between biotic and abiotic drivers, but because they play out over long time-scales they are challenging to resolve empirically. Processbased simulation models as LandClim, when informed and supported by empirical data, have the potential to generate and evaluate hypotheses about the long-term trajectories of such forest systems. Acknowledgements This study was funded by the Federal Ministry for Education and Research (BMBF_NZL_10/019), the Royal Society of New Zealand´s International Mobility Fund (FRG 10-23) and supported by the Bavarian State Ministry of Sciences, Research and the Arts (StMWFK) within the framework of the ‘Bavarian Climate Programme 2020’ in the joint research centre ‘FORKAST’. M. McGlone 57
Manuscript 1: Long-term vegetation dynamics in New Zealand and two anonymous reviewers are gratefully acknowledged for providing helpful comments and suggestions on an earlier version of the manuscript. References Anderson, J. N., Bugmann, H., Dearing, J. and Gaillard, M.-J. 2006. Linking palaeoenvironmental data and models to understand the past and to predict the future. – Trends in Ecology & Evolution 21: 696 704.‐ Becker, R. 2000. Competition in the regeneration niche between conifers and angiosperms: Bond’s slow seedling hypothesis. – Functional Ecology 14: 401 412.‐ Beveridge, A. E. 1973. Regeneration of podocarps in a Central North Island forest. – New Zealand Journal of Forestry 11: 48 55.‐ Bond, W. J. 1989. The tortoise and the hare: Ecology of angiosperm dominance and gymnosperm persistence. – Biological Journal of the Linnean Society 36: 227-249. Brewer, S., Jackson, S. T. and Williams, J. W. 2012. Paleoecoinformatics: Applying geohistorical data to ecological questions. – Trends in Ecology & Evolution 27: 104 112.‐ Briner, S., Elkin, C., Huber, R. and Grêt-Regamey, A. 2012. Assessing the impacts of economic and climate changes on land-use in mountain regions: A spatial dynamic modeling approach. – Agriculture, Ecosystems & Environment 149: 50 63.‐ Brodribb, T. J., Pittermann, J. and Coomes, D. A. 2012. Elegance versus speed: Examining the competition between conifer and angiosperm trees. – International Journal of Plant Sciences 173: 673-694. Bugmann, H. 2001. A review of forest gap models. – Climatic Change 51: 259 305.‐ Bystriakova, N., Bader, M. and Coomes, D. A. 2011. Long-term tree fern dynamics linked to disturbance and shade tolerance. – Journal of Vegetation Science 22: 82 84.‐ Chapin, F. S., Vitousek, P. M. and Cleve, K. van 1986. The nature of nutrient limitation in plantcommunities. – The American Naturalist 127: 48-58. Clarkson, B. D. 1990. A review of vegetation development following recent (450 years) volcanic disturbance in North Island, New Zealand. – New Zealand Journal of Ecology 14: 59 71.‐ Clayton-Greene, K. A. 1977. Structure and origin of Libocedrus bidwillii stands in the Waikato District, New Zealand. – New Zealand Journal of Botany 15: 19 28.‐ Colombaroli, D., Henne, P. D., Kaltenrieder, P., Gobet, E. and Tinner, W. 2010. Species responses to fire, climate and human impact at tree line in the Alps as evidenced by palaeo-environmental records and a dynamic simulation model. – Journal of Ecology 98: 1346-1357. Coomes, D. A., Allen, R. B., Bentley, W. A., Burrows, L. E., Canham, C. D., Fagan, L., Forsyth, D. M., Gaxiola-Alcantar, A., Parfitt, R. L., Ruscoe, W. A., Wardle, D. A., Wilson, D. J. and Wright, E. F. 2005. The hare, the tortoise and the crocodile: The ecology of angiosperm dominance, conifer persistence and fern filtering. – Journal of Ecology 93: 918 935.‐ Dale, V. H. 2005. 25 Years of ecological change at Mount St. Helens. – Science 308: 961 962.‐ Develice, R. L. 1988. Test of a forest dynamics simulator in New Zealand. – New Zealand Journal of Botany 26: 387 392.‐ Druitt, D. G., Enright, N. J. and Ogden, J. 1990. Altitudinal zonation in the mountain forest of Mt. Hauhungatahi, North Island, New Zealand. – Journal of Biogeography 17: 205 220.‐ Ebbett, R. L. and Ogden, J. 1998. Comparative seedling growth of five endemic New Zealand podocarp species under different light regimes. – New Zealand Journal of Botany 36: 189 201.‐ 58
Manuscript 1: Long-term vegetation dynamics in New Zealand Elkin, C., Reineking, B., Bigler, C. and Bugmann, H. 2012. Do small-grain processes matter for landscape scale questions? Sensitivity of a forest landscape model to the formulation of tree growth rate. – Landscape Ecology 27: 697 711.‐ Enright, N. J., Ogden, J. and Rigg, L. S. 1999. Dynamics of forests with Araucariaceae in the western Pacific. – Journal of Vegetation Science 10: 793 804.‐ Grimm, V. and Railsback, S. 2012. Pattern-oriented modelling: A ‘multi-scope’ for predictive systems ecology. – Philosophical Transactions of the Royal Society B: Biological Sciences 367: 298 310.‐ Grimm, V., Revilla, E., Berger, U., Jeltsch, F., Mooij, W., Railsback, S., Thulke, H.-H., Weiner, J., Wiegand, T. and DeAngelis, D. 2005. Pattern-oriented modeling of agent-based complex systems: Lessons from ecology. – Science 310: 987 991.‐ Hall, G. M. J. and Hollinger, D. Y. 2000. Simulating New Zealand forest dynamics with a generalized temperate forest gap model. – Ecological Applications 10: 115. Hall, G. M. J. and McGlone, M. S. 2006. Potential forest cover of New Zealand as determined by an ecosystem process model. – New Zealand Journal of Botany 44: 211 232.‐ Hall, G. M. J., Wiser, S. K., Allen, R. B., Beets, P. N. and Goulding, C. J. 2001. Strategies to estimate national forest carbon stocks from inventory data: The 1990 New Zealand baseline. – Global Change Biology 7: 389-403. Hartig, F., Calabrese, J. M., Reineking, B., Wiegand, T. and Huth, A. 2011. Statistical inference for stochastic simulation models - Theory and application. – Ecology Letters 14: 816 827.‐ Hartig, F., Dyke, J., Hickler, T., Higgins, S. I., O’Hara, R. B., Scheiter, S. and Huth, A. 2012. Connecting dynamic vegetation models to data - An inverse perspective. – Journal of Biogeography 39: 2240-2252. Henne, P. D., Elkin, C. M., Reineking, B., Bugmann, H. and Tinner, W. 2011. Did soil development limit spruce (Picea abies) expansion in the Central Alps during the Holocene? Testing a palaeobotanical hypothesis with a dynamic landscape model. – Journal of Biogeography 38: 933-949. Henne, P. D., Elkin, C., Colombaroli, D., Samartin, S., Bugmann, H., Heiri, O. and Tinner, W. 2012. Impacts of changing climate and land use on vegetation dynamics in a Mediterranean ecosystem: Insights from paleoecology and dynamic modeling. – Landscape Ecology. Hereford, J. 2009. A quantitative survey of local adaptation and fitness trade offs. – The American ‐ Naturalist 173: 579-588. Hogg, A., Lowe, D. J., Palmer, J., Boswijk, G. and Ramsey, C. B. 2012. Revised calendar date for the Taupo eruption derived by 14C wiggle-matching using a New Zealand kauri 14C calibration data set. – The Holocene 22: 439 449.‐ Horrocks, M. and Ogden, J. 1998. The effect of the Taupo tephra eruption of c. 1718 BP on the vegetation of Mt. Hauhungatahi, central North Island, New Zealand. – Journal of Biogeography 25: 649 660.‐ Kunstler, G., Coomes, D. A. and Canham, C. D. 2009. Size-dependence of growth and mortality influence the shade tolerance of trees in a lowland temperate rain forest. – Journal of Ecology 97: 685 695.‐ Land Information New Zealand, L. 2011. Digital elevation map of the North Island, DEM25. Landcare Research 2011a. Ecological Traits of New Zealand Flora. – Ecological Traits of New Zealand Flora – http://ecotraits.landcareresearch.co.nz/WeedsIntro.aspx. 59
Manuscript 1: Long-term vegetation dynamics in New Zealand Landcare Research 2011b. Ngā Tipu o Aotearoa - New Zealand Plant Names Database. – Ngā Tipu o Aotearoa - New Zealand Plant Names Database – http://nzflora.landcareresearch.co.nz/. Leathwick, J. R. 1995. Climatic relationships of some New Zealand forest tree species. – Journal of Vegetation Science 6: 237 248.‐ Leathwick, J. R. and Mitchell, N. D. 1992. Forest pattern, climate and vulcanism in Central North Island, New Zealand. – Journal of Vegetation Science 3: 603 616.‐ Lecointre, J. A., Neall, V. E., Wallace, R. C., Elliot, M. B. and Sparks, R. 2004. Late Quaternary evolution of the Rotoaira Basin, northern Tongariro ring plain, New Zealand. – New Zealand Journal of Geology and Geophysics 47: 549 565.‐ Loehle, C. 1998. Height growth rate tradeoffs determine northern and southern range limits for trees. – Journal of Biogeography 25: 735-742. Lusk, C. H. and Matus, F. 2000. Juvenile tree growth rates and species sorting on fine-scale soil fertility gradients in a Chilean temperate rain forest. – Journal of Biogeography 27: 1011-1020. Lusk, C. H. and Ogden, J. 1992. Age structure and dynamics of a podocarp-broadleaf forest in Tongariro National Park, New Zealand. – Journal of Ecology 80: 378-393. Lusk, C. H. and Pozo, A. del 2002. Survival and growth of seedlings of 12 Chilean rainforest trees in two light environments: Gas exchange and biomass distribution correlates. – Austral Ecology 27: 173-182. Lusk, C. H., Duncan, R. P. and Bellingham P. J. 2009. Light environments occupied by conifer and angiosperm seedlings in a New Zealand podocarp-broadleaved forest. – New Zealand Journal of Ecology 33: 83 89.‐ Martin, T. J. and Ogden, J. 2006. Wind damage and response in New Zealand forests: A review. – New Zealand Journal of Ecology 30: 295-310. McGlone, M. S. 1989. The Polynesian settlement of New Zealand in relation to environmental and biotic change. – New Zealand Journal of Ecology: 115-129. McGlone, M. S. and Moar, N. T. 1977. The Ascarina decline and post-glacial climatic change in New Zealand. – New Zealand Journal of Botany 15: 485 489.‐ McGlone, M. S., Hall, G. M. J. and Wilmshurst, J. M. 2011. Seasonality in the early Holocene: Extending fossil-based estimates with a forest ecosystem process model. – The Holocene 21: 517 526.‐ McGlone, M. S., Neall, V. E. and Clarkson, B. D. 1988. The effect of recent volcanic events and climatic changes on the vegetation of Mt Egmont (Mt Taranaki), New Zealand. – New Zealand Journal of Botany 26: 123 144.‐ McGlone, M. S., Richardson, S. J. and Jordan, G. J. 2010. Comparative biogeography of New Zealand trees: Species richness, height, leaf traits and range sizes. – New Zealand Journal of Ecology 34: 137-151. McKelvey, P. J. 1963. The synecology of the West Taupo indigenous forest. – Bulletin of the New Zealand Forest Service 14: 1 127.‐ Mladenoff, D. J. 2004. LANDIS and forest landscape models. – Ecological Modelling 180: 7-19. Niinemets, Ü. and Valladares, F. 2006. Tolerance to shade, drought, and water logging of temperate northern hemisphere trees and shrubs. – Ecological Monographs 76: 521-547. NIWA 2011. CliFlo NIWA’s National Climate Database – http://cliflo.niwa.co.nz/. Ogden, J. and Stewart, G. 1995. Community dynamics of the New Zealand conifers. – In: Enright, N. and Hill, R. (eds.), Ecology of the Southern Conifers. Melbourne University Press, pp. 81‐ 119. 60
Manuscript 1: Long-term vegetation dynamics in New Zealand Ogden, J., Braggins, J., Stretton, K. and Anderson, S. 1997a. Plant species richness under Pinus radiata stands on the Central North Island volcanic plateau, New Zealand. – New Zealand Journal of Ecology 21: 17 29.‐ Ogden, J., Fordham, R. A., Horrocks, M., Pilkington, S. and Serra, R. G. 2005. Long-term dynamics of the long-lived conifer Libocedrus bidwillii after a volcanic eruption 2000 years ago. – Journal of Vegetation Science 16: 321 330.‐ Ogden, J., Horrocks, M., Palmer, J. G. and Fordham, R. A. 1997b. Structure and composition of the subalpine forest on Mount Hauhungatahi, North Island, New Zealand, during the Holocene. – The Holocene 7: 13 23.‐ Palmer, J. G. and Xiong, L. 2004. New Zealand climate over the last 500 years reconstructed from Libocedrus bidwilllii hook. F. tree-ring chronologies. – The Holocene 14: 282 289.‐ Perry, G. L. W. and Millington, J. D. A. 2008. Spatial modelling of succession-disturbance dynamics in forest ecosystems: Concepts and examples. – Perspectives in Plant Ecology, Evolution and Systematics 9: 191 210.‐ R Development Core Team 2012. R: A language and environment for statistical computing, reference index version 2.15.2 – R Foundation for Statistical Computing – www.R-project.org. 2012-12-09. Rogers, G. M. and McGlone, M. S. 1989. A post-glacial vegetation history of the southern-central uplands of North Island, New Zealand. – Journal of the Royal Society of New Zealand 19: 229‐ 248. Scheffer, F. and Schachtschabel, P. 2002. Lehrbuch der Bodenkunde – Spektrum Akademischer Verlag GmbH. Schumacher, S. 2004. The role of large-scale disturbances and climate for the dynamics of forested landscapes in the European Alps. Schumacher, S. and Bugmann, H. 2006. The relative importance of climatic effects, wildfires and management for future forest landscape dynamics in the Swiss Alps. – Global Change Biology 12: 1435-1450. Schumacher, S., Bugmann, H. and Mladenoff, D. 2004. Improving the formulation of tree growth and succession in a spatially explicit landscape model. – Ecological Modelling 180: 175 194.‐ Smale, M. C. and Smale, P. N. 2003. Dynamics of upland conifer/broadleaved forest at Waihaha, Central North Island, New Zealand. – Journal of the Royal Society of New Zealand 33: 509‐ 528. Smith, T. and Huston, M. 1989. A theory of the spatial and temporal dynamics of plant communities. – Vegetatio 83: 49-69. Strigul, N., Pristinski, D., Purves, D., Dushoff, J. and Pacala, S. 2008. Scaling from trees to forests: Tractable macroscopic equations for forest dynamics. – Ecological Monographs 78: 523-545. Uriarte, M., Canham, C. D., Thompson, J., Zimmerman, J. K., Murphy, L., Sabat, A. M., Fetcher, N. and Haines, B. L. 2009. Natural disturbance and human land use as determinants of tropical forest dynamics: Results from a forest simulator. – Ecological Monographs 79: 423 443.‐ Valladares, F. and Niinemets, Ü. 2008. Shade tolerance, a key plant feature of complex nature and consequences. – Annual Review of Ecology, Evolution, and Systematics 39: 237-257. Veblen, T. T. and Stewart, G. H. 1982. On the conifer regeneration gaps in New Zealand: The dynamics of Libocedrus bidwillii stands on the South Island. – Journal of Ecology 70: 413 436.‐ 61
Manuscript 1: Long-term vegetation dynamics in New Zealand Wardle, P. 1964. Facets of the distribution of forest vegetation in New Zealand. – New Zealand Journal of Botany 2: 352 366.‐ Wells, A., Duncan, R. P. and Stewart, G. H. 2001. Forest dynamics in Westland, New Zealand: The importance of large, infrequent earthquake-induced disturbance. – Journal of Ecology 89: 1006 1018.‐ Wilmshurst, J. M. and McGlone, M. S. 1996. Forest disturbance in the Central North Island, New Zealand, following the 1850 BP Taupo eruption. – The Holocene 6: 399 411.‐ Wilson, J. B. and Lee, W. G. 2012. Is New Zealand vegetation really ‘problematic’? Dansereau’s puzzles revisited. – Biological Reviews 87: 367 389.‐ Supplement S1: LandClim – model description Detailed descriptions of the full structure of LandClim are given in Schumacher (2004), Schumacher et al. (2004) and Schumacher and Bugmann (2006). In order to explain the effect of biotic and abiotic drivers of vegetation dynamics, the following section provides a brief overview of the stand-level processes of LandClim. The establishment, growth and mortality of trees are the three fundamental stand-scale processes considered by LandClim. The model tracks individual trees in the aggregated form of cohorts described by their biomass. Cohorts are groups of same-aged trees within a grid-cell (25 × 25 m) with all individuals having the same biomass (which can be translated into diameter at breast height using the allometric given below in section ‘Growth’) and survival probability, but with mortality occurring at the individual-level. Establishment For the representation of establishment, tree species are filtered by propagule availability and physical properties within each cell. Establishment in a cell is possible only in years where the light available at the forest floor exceeds a speciesspecific threshold value, winter temperature (mean temperature of the coldest month) is higher than a species-specific minimum temperature, and the sum of growing degree-days exceeds the species-specific minimum requirement. A species-specific establishment coefficient affects the probability of actual establishment. To reduce computational effort, the number and biomass of trees in any newly established cohort are calculated decadally. 62
Manuscript 1: Long-term vegetation dynamics in New Zealand Growth Maximum growth rates are assigned to each species (representing the growthrate under optimum environmental conditions), which are reduced by light availability, sum of degree days and drought. Actual tree growth is determined by a combination of these reduction factors under the application of Liebig's ‘Law of the minimum’. For conversion of the model biomass output into DBH, an allometric relationship was derived from individual-level data for Weinmannia racemosa (kamāhi) and Dacrydium cupressinum (rimu), reported by Beets et al. (2008): d=45.59799⋅b0.39877240 Eq. S1 where: d = diameter at breast height (cm), and b = aboveground biomass (kg; assuming that 50% of dry trunk biomass is carbon). Mortality Tree mortality arises from three broad factors: 1) growth-dependent stress, 2) density-dependent stress and 3) an intrinsic, age-related component. Growthdependent stress occurs when the annual growth rate drops below a speciesspecific threshold value due to unfavourable environmental conditions. Stressrelated mortality probability increases after a minimum number of consecutive low-growth years accumulate. Density-dependent mortality occurs only if total stand biomass exceeds the maximum stand biomass in a given cell. Agedependent mortality increases with tree age. The probability of an individual tree dying is given by the maximum of these three probabilities. New Zealand’s temperate forests can attain substantial amounts of standing biomass per area (Wardle 1991), exceeding those of central European forests. Therefore, the maximum stand biomass was increased from 300 to 1000 t/ha. This increased limit to standing biomass effectively reduces the probability of densitydependent mortality. S2: Parameterization of tree species life-history traits maxAge The parameter ‘maxAge’ characterises the expected longevity of a tree species (Bugmann 1994, Schumacher et al. 2004). For the Hauhungatahi species, the max- 63
Manuscript 1: Long-term vegetation dynamics in New Zealand imum age was assigned based on the age-estimates of oldest individuals given by the following studies: Ogden et al. (2005) for L. bidwillii (1000 years), Smale and Smale (2003) for P. unninghamii (650 years), Lusk and Ogden (1992) for W. racemosa (400 years). A value of 800 years was assigned to D. cupressinum since a wide range of maximum ages have been reported (500 up to 1000 years, Lusk and Ogden 1992). An estimated maximum age of 150 years was assigned to the tree fern life form. Shade tolerance Species shade tolerance expressed in classes between 1 and 5, with 1 denoting the least shade tolerance (following the classification given by Ellenberg 2009). These classes define species-specific minimum light requirements (see Schumacher, 2004, p.122, Tab. B.3). Shade tolerance classes for the tree species of Mt. Hauhungatahi were determined via the pattern-oriented parametrization approach. Rmax The parameter ‘Rmax’ determines the maximum above-ground biomass growth rate per year, and is expressed relative to the maximum biomass of a tree species. The ranges for ‘Rmax’ were determined by the pattern-oriented parameterization approach and a refined sensitivity analysis. Average growth rates were determined by a linear regression of the age-DBH data for both empirical and model data. The empirical sources for growth data were: Ogden et al. (2005) for L. bidwillii (mean annual growth rate of 0.125 cm/yr), Lusk and Ogden (1992) for W. racemosa (0.209 cm/yr) and D. cupressinum (0.124 cm/yr), Smale and Smale (2003) for P. cunninghamii (0.15 cm/yr). Kmax Maximum biomass of an individual tree (‘Kmax’) was estimated from an allometric relationship given in Hall et al. (2001), relating DBH and height to biomass. Unless otherwise noted, values of mean maximum tree height were taken from the New Zealand Eco-Trait database (Landcare Research 2011a), maximum DBH-values were taken from the electronic version of the Flora of New Zealand (Landcare Research 2011b). For P. cunninghamii, the DBH value reported in Russo et al.´s (2007) Tab. S1 was used. For D. cupressinum, both DBH and height values were derived from the Flora of New Zealand (Landcare Research 2011b). 64
Manuscript 1: Long-term vegetation dynamics in New Zealand Since several maximum heights were reported, the standard maximum height listed in the Eco-Traits database was assumed and the resulting ‘Kmax’ value rounded to the nearest whole number. For L. bidwillii, the DBH and height values given in the Flora and Eco-Trait databases (Landcare Research 2011a, Landcare Research 2011b) resulted in a ‘Kmax’ value of 3.92. Since L. bidwilli did not occur in the carbon stock assessment report of Beets et al. (2008), we used the maximum DBH values given by Ogden et al. (2005) as calibration targets and found that with Kmax = 3.92 LandClim significantly under-predicted DBH (with a maximum around 60 cm compared to the 100+ cm observed by Ogden et al. 2005). Therefore, ‘Kmax’ was increased to 8, which resulted in a more reasonable maximum DBH of 90-100 cm. Preliminary test scenarios using a wide range of ‘Kmax’ values for L. bidwillii showed that the model was reasonably robust to changes in this parameter’s value. Leaf Habit The ‘LeafHabit’ distinguishes between ‘Evergreen’, ‘Broadleaved evergreen’ and ‘Deciduous’ species (Schumacher 2004). Classes were assigned according to leaf morphology information given on the NZ Eco-Trait database (Landcare Research 2011a). Foliage type Foliage type is a parameter ranked in five classes that determines the shading potential of a tree species. An allometric function is applied which scales from foliage characteristics to the shade cast by a tree individual (via tree size expressed as DBH).We estimated the foliage type based on information about leaf morphology given in the Eco-Trait database (Landcare Research 2011a) and from expert opinion about the species canopy structure and shading potential. Min Temperature This parameter determines the minimum temperature (°C) for species establishment (Bugmann 1994). We used the freezing resistance reported by Sakai and Wardle (1978) to assign this trait to L. bidwillii, D. cupressinum, P. cunninghamii and W. racemosa. For the tree fern life-form the same minimum temperature (-7 °C) as W. racemosa was assumed. 65
Manuscript 2: Climatic turning point for beech and oak is strong in operational forest management planning), LandClim simulates longterm dynamics of forest at the landscape scale considering particularly demographic processes such as establishment, growth and mortality. To determine the climatic turning point (CTP) for a mixed beech-oak forest of considerable relevance in Central Europe we combine empirical evidence and modelling. Data from repeated forest surveys and tree-ring analysis are used to assess current tree growth and to harmonize the forest models. These are then used to predict the forest development under climate change. This approach 72 Tab. 1: A systematic literature search (ISI Web of Science ® Data base) on growth of beech and oak under drought resulted in 20 papers dealing with the coexistence of beech (Fagus sylvatica) and oak (Quercus petraea and/or Q. robur) (cf. Literature review S2). Notably, only two studies examine competition dynamics among the two species under future climates. Method: M=Modeling F=Field observation E=Experiment. Reference Method Target period Target driver Aim Bellassen et al. 2011 M past no model validation Bonn 2000 F past climate/weather tree growth and competition Bugmann and Cramer 1998 M past soil water balance tree growth and competition Czúcz et al. 2011 M past climate tree growth and competition Davi et al. 2009 M past carbon and water fluxes tree growth and competition Fabbio et al. 2006 F past forest management/extreme event tree growth and competition Friedrichs et al. 2009 F past climate tree growth García-Suárez et al. 2009 F past climate tree growth and competition Granier et al. 2007 M past climate/weather/extreme event tree growth Hlásny et al. 2011 M future climate tree growth Leuschner et al. 2001 F past soil water balance tree growth and competition Mérian & Lebourgeois 2011 F past climate tree growth Piedallu et al. 2011 F past soil water balance tree growth, modelling, soil water holding capacity Pollastrini et al. 2010 E future climate tree growth Scharnweber et al. 2011 M past climate tree growth and competition Scherrer et al. 2011 F past climate/extreme event tree growth and competition Solberg et al. 2009 F past climate/soil chemistry tree growth Szabados 2006 F past climate tree growth van der Werf et al. 2007 F past extreme event tree growth Zang et al. 2011 F past climate/extreme event tree growth and competition
Manuscript 2: Climatic turning point for beech and oak allows us to perform a straightforward analysis of the following two decisive hypotheses: 1) Projected future climate eventually crosses a climatic turning point with a drought-induced rank reversal in the species dominance and 2) the rank reversal is driven not only by climate, but also by the inherited stand structure. Whilst literature is rich on climatic limits for species existence, we are concerned with the gradual response of ecosystem shifts in species composition within these limits and work out the CTP which serves as a landmark for investigating changing interspecific competition. A very cautious approach allows us to trace and discuss uncertainties in the determination of the CTP and their consequences for forest management practice. Materials and Methods Study site and data The studied beech-oak forest stands are located in south-eastern Germany on the ‘Franconian plateau’ in an altitude of 300-400 m a.s.l.. Four sites are permanent investigation sites of different stand age which belong to the Bavarian long-term experimental plot network, the fifth site is a forest reserve of the Bavarian state forest institute LWF (Tab. 2). The experimental plots were installed in 73 Tab. 2: Stand parameters from surveys of the studied beech-oak stands near Schweinfurt. N = Stem number per hectare, H100 = Upper canopy/ top height, G = Basal area, V = Merchantable wood volume, dg = Root of mean diameter squared, B = Aboveground biomass (Schroeder et al. 1997) Stand (initial age) Survey year All tree species Beech Oak Other N (ha-1) H100 (m) G (m2/ha) V (m3/ha) B (t/ ha) dg (cm ) G (m2/ha) dg (cm ) G (m2/ha) dg (cm ) G (m2/ha) 27yrs 1995 3160 15.7 25.2 143.6 536 11.4 6.4 9.6 17.0 11.3 1.8 2005 1751 19.9 28.4 233.2 824 15.1 9.1 14.3 18.0 11.4 1.3 54yrs 1995 1935 22.1 34.4 311.4 715 15.7 15.2 19.1 9.9 12.1 9.3 2005 1384 25.2 40.7 449.3 778 20.4 19.3 22.5 12.1 15.6 9.3 86yrs 1995 1149 24.3 30.1 316.8 561 15.4 11.2 22.7 15.4 16.3 3.5 2005 1037 26.8 36.9 443.2 636 18.5 14.2 25.7 18.8 18.6 3.9 106yrs 1995 676 27.8 25.7 323.3 422 18.8 14.4 30.9 10.4 24.9 0.9 2005 615 30.4 30.5 425.7 485 21.4 16.6 34.9 12.8 27.1 1.1 Reserve 1978 723 25.8 30.4 355.1 472 27.2 12.9 35.6 6.4 18.0 11.1 (105 yrs) 1996 531 30.5 37.0 528.1 644 34.6 18.5 42.8 7.5 22.3 11.0 2010 396 33.8 38.6 622.4 593 40.7 22.4 44.5 6.5 26.0 9.7
Manuscript 2: Climatic turning point for beech and oak 1995 and surveyed in 1995 and 2005. The stands are 2.25-6.8 ha large, even-aged and range from 27-106 years with merchantable wood volumes of 233-323 m3/ha in 1995. The basal area ratios of oak:beech:other species range from 25:67:7 to 56:40:4, hornbeam (Carpinus betulus L.) shares up to 32%, lime (Tilia cordata P. Mill.) and maple (Acer pseudoplatanus L.) cover minor percentages. No thinning occurred between 1995 and 2005. The forest reserve with a core area of 0.99 ha was designated in 1978 and surveyed in 1978, 1996 and 2010. Since that time any forest management is forbidden. In 1978, tree age in the core area was 105 years, wood volume was 311 m³/ha and basal area ratio of oak:beech:other species was 42:21:37 (Tab. 2). For retrospective growth analyses, increment cores were extracted in the direct vicinity of the 86-year experimental plot in 2009. Two cores of each 15 dominant oak and beech trees were sampled at breast height. After wood surface preparation, ring width of each sample was measured using a LINTAB linear table (Rinn 2003) to a precision of 0.01 mm. The software TSAP-Win (Rinn 2003) was used to synchronize the ring width curves visually and statistically. The biological age trend in the original tree-ring series was eliminated using the dplR library v 1.5.5 (Bunn 2008) in R v 2.15.1 (R Development Core Team 2012) by applying a cubic smoothing spline of 2/3 of the series length. Autoregressive modelling was used to remove first order autocorrelation. Residual chronologies of beech and oak were obtained by averaging the ring width series using a biweight robust mean (Cook and Peters 1997). Climate data Climate data (daily values of temperature and precipitation for the period 01.05.1958 - 31.12.2003) were obtained from the weather station ‘Schweinfurt- Gartenstadt’ close to the study site (DWD station 4621, 50.06 °N; 10.22 °E; 240 m a.s.l.). These data were used for the retrospective growth analysis and calibration of SILVA and LandClim. Basis for the projections of possible future forest development was the WETTREG 2010 A1B ‘normal’ scenario for the same climate station from 1961-2100 (Kreienkamp et al. 2009). For this scenario, ten realizations, with each of them covering two decades, are available. These climate time series provide a climatic gradient from moderate temperature and soil moisture conditions in the 74
Manuscript 2: Climatic turning point for beech and oak past and present to warmer and dryer conditions in the future, including a realistic temporal variability and autocorrelation (Tab. S1). Compared to 1971-2000 the WETTREG scenario projects for 2071-2100 a significant increase in the mean annual and May-Sep temperatures from 9.0 °C to 12.5 °C and 15.9 °C to 19.4 °C. The growing season with mean temperatures above 10 °C is thereby extended from 164 to 202 days (Fig. 1). Annual precipitation decreases from 558 mm to 523 mm and summer precipitation from 258 to 200 mm. Current and future stand dynamics In a first step, the empirical data were used to analyse tree growth and competition between beech and oak under past and present climatic conditions. In the present study a species’ dominance was defined by its basal area (SILVA) or aboveground biomass (LandClim) being higher than any other species’ basal area/biomass. A species’ competitiveness was defined in terms of a gain or loss in its basal area share or biomass share. Competitiveness (and ultimately dominance) is determined by the conjoint effect of the demographic processes establishment, growth, and mortality. Basal area and biomass are standard measures in mixed species forest ecology (Pretzsch and Schütze 2009). 75 Fig. 1: Present and future climate of the study site Schweinfurt (south-eastern Germany). a) Present climate for the period 1971-2000 according to the DWD climate station Schweinfurt, b) future climate from 2071-2100 according to the WETTREG 2010 A1B scenario. As in Walter-Lieth climate diagrams hatched area indicate moist periods, dotted areas dry periods (Walter and Lieth 1967).
Manuscript 2: Climatic turning point for beech and oak In a second step, the forest growth models SILVA and LandClim were employed to assess future forest dynamics, particularly the ‘climatic turning point’ (CTP). We define a climatic turning point as the climate where a rank reversal of the species’ dominance, i.e. basal area share (SILVA) and biomass share (LandClim) occurs. As postulated, the rank reversal and thereby the observed CTP might be affected by inherited stand structure. By ‘inherited stand structure’ we refer to the footprint of prior climate conditions and forest management causing a gradual response of forest structure and species composition in long-lived ecosystems with long generation periods. Therefore, a potential and actual CTP was distinguished. To determine the potential CTP forest succession was simulated for forests developing ex novo, i.e. establishing from seed rain, under different, but stationary climate conditions derived from the WETTREG data. The actual CTP was determined by the simulation of a forest succession in present forests along the projected gradient from moderate temperature and soil moisture to warmer and dryer conditions (WETTREG time series). The models SILVA is a site-condition sensitive, single-tree based, spatially explicit forest growth model (Pretzsch et al. 2002). The growth functions were statistically parameterized from data of long-term experimental plots across Central Europe. The parameterizations for beech and oak were based on 13,000 trees and 3,000 trees of altogether 89+37 plots, respectively. SILVA was developed as a growth simulator for operational forest management planning of the Bavarian state forest management, for educational and scientific purposes (Pretzsch et al. 2008). Due to its sitesensitivity, SILVA simulations have been included in the analysis of climate change effects on forest growth and the development of forest carbon stocks under different climate and management scenarios (Köhl et al. 2010, Rötzer et al. 2010). LandClim is a spatially explicit forest landscape model that was developed to assess the importance of climatic effects, wildfire and management on forest dynamics (Schumacher et al. 2004, Schumacher and Bugmann 2006, Schumacher et al. 2006). It incorporates patch level processes of climate-dependent tree regeneration, growth and mortality, and landscape level processes such as forest disturbances and seed dispersal. Thereby several stand generations can be simulated and the influence of variables changing with time or space on forest dynamics can be assessed. LandClim operates on long time scales (hundreds to thousands of 76
Manuscript 2: Climatic turning point for beech and oak years) and large spatial extents (several hectares) at a relatively fine scale (grid cells of 25 by 25 m, monthly weather data), and has been tested and adapted to the European Alps, the North American Rocky Mountains, and Mediterranean forests (Schumacher and Bugmann 2006, Colombaroli et al. 2010, Henne et al. 2011, Briner et al. 2012, Elkin et al. 2012, Henne et al. 2012). Species parameters were taken from Schumacher et al. (2004) and Henne et al. (2011) except for an increase in the drought tolerance of Quercus petraea from 3.0 to 3.5 to reflect the slightly higher drought tolerance than Fagus sylvatica (drought tolerance = 3.0) reported in the literature (e.g. Leuschner et al. 2001, Raftoyannis and Radoglou 2002, Bréda et al. 2006, Scharnweber et al. 2011). For model validation the SILVA simulations which were initialised with the experimental plot data could be compared directly with the true stand development. The simulation results for the 1981-2020 periods were validated against the data from the experimental plots and forest reserve. The simulated basal area growth matched the actual basal area well, except for 10-15% underestimation in the case of the 54 and 86-yrs old stands. The difference was due to an underestimation of the increment of the ‘other species’ which cluster several species with different growth characteristics. The growth ratio of oak to beech was generally estimated correctly; oak increment was slightly overestimated in the 27-yrs old stand. These results are acceptable for our purposes; a perfect match of reality is unrealistic since the stand interactions with the environment are much too complex for a stable model: E.g. late frost or other climate hazards, insect calamities, mast years all have a very unpredictable pattern and are essentially smoothed in the SILVA parameterization (Fig. S1). LandClim was validated by translating the permanent plot data into biomass using the allometric function implemented in LandClim (Schroeder et al. 1997). Simulations were run with the current climate (DWD data) and results were compared to the real stands. As LandClim does not take into account the anterior oak fostering, stand biomass is dominated by beech which is more competitive under the current climate (Fig. S2). 77
Manuscript 2: Climatic turning point for beech and oak Simulations Mixed beech-oak forest growth under climate conditions from the WETTREG 2010 A1B scenario was simulated in order to determine the CTP between both species (Tab. 3). For the potential CTP, climate had to be stationary during the simulated succession so that forests developed in equilibrium with climate. To create such climate scenarios the WETTREG 2010 A1B time-series covering the period 1961-2100 was split into 20-year periods (for each of ten available realizations). Seven climate scenarios were generated by looping each 20-year time- 78 Tab. 3: Definition of simulation runs. Temperature and precipitation are given per annum (Temp.a, Prec.a) and per growing season (Temp.v, Prec.v), i.e. May-September of a respective year. Simulation Model Initial age (yrs) Sim. period (yrs) Final age (yrs) WETTREG A1B period (yrs) Temp.a (Temp.v) (°C) Prec.a (Prec.v) (mm) Potential climatic turning point (simulation with quasiconstant climate conditions) SILVA 27 (experimental stand) 120 147 1961-1980 (CLIM1) 8.9 (15.6) 546 (256) 1981-2000 (CLIM2) 8.9 (15.9) 553 (251) 2001-2020 (CLIM3) 9.3 (16.4) 561 (245) LandClim 0 (establishment from ‘global seed rain’) 150 150 2021-2040 (CLIM4) 10.3 (17.2) 534 (219) 2041-2060 (CLIM5) 11.1 (18.2) 512 (227) 2061-2080 (CLIM6) 11.9 (19.0) 500 (200) 2081-2100 (CLIM7) 12.6 (19.4) 516 (195) Simulation span Data Actual climatic turning point (continuous simulation acc. to WETTREG scenario) SILVA 27 (experimental stand) 100 127 2001-2100 WETTREG A1B time series 54 (experimental stand) 100 154 86 (experimental stand) 100 186 106 (experimental stand) 100 206 105 (forest reserve) 120 205 1981-2100 WETTREG A1B time series LandClim 0 (establishment from ‘global seed rain’) 180 180 1921-2100 From 1921 to 1960 WETTREG A1B time series ‘1961-1980’ was repeated twice, afterwards WETTREG A1B time series
Manuscript 2: Climatic turning point for beech and oak series as often as needed (for 150 years succession 7.5 times). Forest structure and species composition were evaluated at a stand age of 150 years. This is the upper limit for beech rotation and lower limit for oak rotation. Neither species should suffer from an age-dependent mortality. According to the different initialisation requirements of the growth models SILVA and LandClim, SILVA was initialised with the 27-year old experimental stand (Tab. 2) and LandClim with a beech-oak ‘global seed rain’ on bare ground. For the actual CTP, beech-oak stand development was simulated according to the predicted climate development from the WETTREG 2010 A1B scenario until 2100 (for each of ten available realizations). Again, SILVA and LandClim had to be initialised differently. SILVA simulated stand dynamics of the four experimental plots and the forest reserve (Tab. 2). LandClim generated a hypothetical stand which was in 1961 already 40 years old. Following the initial beech-oak global seed rain, the first 40 years were assumed to grow under climate conditions similar to those from 1961 to 1980. In 2000, the LandClim stand was approximately as old as the 106-year-old experimental plot stand and the forest reserve (105 yrs). Although climate data are identical, SILVA and LandClim use different data aggregations and time windows. For SILVA the temperature and precipitation data were translated into length of growing season, average temperature May-Sep, temperature amplitude Jan-Jul, precipitation May-Sep and the Martonne aridity index, each of the parameters averaged over 20 year periods (Pretzsch 2009). In contrast, LandClim worked with monthly mean temperatures and precipitation sums to calculate seasonal and annual indices for water availability (drought index; Bugmann and Cramer 1998, Bugmann and Solomon 2000) and energy availability (degree-day-sum; Bugmann 1994) as well as temperature of the coldest month. Beside differences in the representation of climate both models made specific assumptions on soil water and nutrient availability. Soil descriptions for the experimental plots were translated into the model requirements. Specifically, for the SILVA simulations a low soil water availability was assumed (0.3 on a scale between 0…1) and further dropped as the climate becomes warmer and drier. For LandClim, the maximum available soil water capacity, represented in the model as buckets size, was set to 80 mm. 79
Manuscript 2: Climatic turning point for beech and oak Results Current stand dynamics In all but the 106 years old stands, beech expanded its basal area while oak remained unchanged or declined (Fig. 2). Beech expansion was particularly strong in the young 27 and 86 years old stands and in the forest reserve which had the longest time period between surveys. Here, from 1978 to 2010, the beech basal area share increased from 42% to 55%, partly at the expense of oak (decline from 21% to 17%), but mainly due to the decline of other species (37% to 25%). A closer look at the 86-yrs old experimental site emphasised the dependence of basal area growth response on stand structure (DBH-distributions in Fig. S3). The tree-ring analysis showed an increase of mean DBH of dominant beech from 31.6 to 37.5 cm (+ 5.9 cm) and oak from 26.0 to 29.8 cm (+ 3.8 cm). So, in terms of diameter increase, beech trees exceeded oak trees by 40-60% in the upper canopy. However, the majority of the beech trees still grew in the subcanopy and was strongly suppressed so that, in total, the basal area share remained unchanged (Fig. S3). 80 Fig. 2: Basal area share of beech, oak and other tree species of the four experimental plots (in 1995 and 2005) and the forest reserve (in 1978, 1996, and 2010) according to field surveys.
Manuscript 2: Climatic turning point for beech and oak Correlations between tree-ring indices and climate data revealed that beech was much more sensitive to climate than oak. While oak only showed significant positive correlations with precipitation of the growing period, beech tree-ring indices were positively correlated with precipitation during spring and the growing season as well as to September during the year prior to growth. Correlations with temperature were negative for most months except for October of the previous year (Fig. S4). In the driest summer of 2003 (100 mm precipitation from May-Sep), beech ring-width index dropped to 0.5 while that for oak only decreased to 0.7 (Fig. 3a). However, the absolute ring-width values (Fig. 3b) showed that oak still did not grow stronger than beech in that summer (both species grew 1.2 mm). A 2nd degree polynomial (in approximation of an optimum-function) fitted to the treering data had a stronger curvature for beech than for oak (Fig. 3). The hypothetical intersection between the two polynomials lied even below 100 mm precipitation (May-Sep). Thus, under current climate conditions, the tree-ring data confirmed a stronger competitiveness of beech than oak, even in years of extremely low precipitation. Future stand dynamics Potential climatic turning point To search the potential CTP between beech and oak, the development of beech-oak stands under constant climate conditions calculated from seven consecutive 20 year time periods between 1961 and 2100 (WETTREG 2010 A1B) was simulated. The results of the SILVA simulations indicated that in all but the CLIM7 scenario (2081-2100) beech extended its initially low basal area share (Fig. 4a). However, the beech expansion decreased along the climatic gradient. In the last two scenarios CLIM6-7 (2061-2100) beech did not dominate anymore. While in CLIM6 (2061-2080) beech still slowly expanded and might eventually reach dominance in the long-term, in CLIM7 (2081-2100) beech basal area did not further expand. Notably, not only species composition changed, but also the final basal area value decreased from almost 38 m2/ha in CLIM1 (1961-1980) to 26 m2/ha in CLIM7 (2081-2100) – the latter with no net basal area increase compared to the initial 27 yr old stand. 81
Manuscript 2: Climatic turning point for beech and oak Controversially discussed but not considered in the present study is the beech- -oak suitability with respect to soil texture, fertility and water relations. Traditionally, oak is regarded as more competitive on extremely clayey, acidic-nutrient poor or water stagnant soils (Ellenberg 2009). The evaluation of natural beech invasion in unmanaged 50 years old seeded oak stands in south-western Germany showed that on limestone oak was pushed back to 5-25 %, while on sandstone beech only conquered 5-25 % (pers. comm. Abt 2012). Compared to our sites, we suggest that on more oak favourable soils, beech might – if at all – only slowly (perhaps after generations) become dominant. Examples of a re-conquering of oak-forests through beech even on extremely poor soils are known (Ellenberg 2009). Implications for management Arguably like no other life science, forest management is forced to think ahead of time due to its long production time spans. For more than 30 years, forest management in Central Europe has pursued a more sustainable and stable productivity through conversion of not site-suited coniferous forest to broadleaf forest and propagation of species mixtures (Spiecker et al. 2004, Pretzsch et al. 2013). As climate change moved into the focus of attention, species suitability decisions had to take into account possible future climate trends (Linder 2000, Bolte et al. 2009, Lindner et al. 2010). The important pending decision for already warmer and drier (edaphically not extreme) sites in Central Europe is beech and/or oak, and threshold values are sought for orientation. Temperature and precipitation are the most commonly applied factors and our study demonstrates that beech is currently not drought-limited in south-eastern Germany, even at the most dry sites. Simulations with SILVA and LandClim similarly predict limits of beech dominance for temperatures of 11.1 °C (annually), 18.2 °C (May-Sep), 20.9 °C (July), and precipitation of 510 mm (annually) and 230 mm (May-Sep). These values remain within the climate envelopes of Kölling (2007) for beech existence, but fall below the typical values given for beech dominance (cf. Bolte et al. 2007). Based on these values, there is still a buffer for beech forests also on the dryer sites in southeastern Germany – at least until 2050 even under ‘worst-case’ assumptions like the WETTREG 2010 A1B scenario. However, in view of the predicted beech decline in both SILVA and LandClim in the second half of the 21st century, the promotion of oak already today seems 88
Manuscript 2: Climatic turning point for beech and oak advisable due to the superior resistance and resilience of oak (pedunculate oak even more than sessile) to drought (Leuschner et al. 2001, Friedrichs et al. 2009, Scharnweber et al. 2011). Also the high value of oak wood might make oak more profitable than beech if stand productivity decreases as projected. Additionally, the facilitative behaviour of oak in mixed stands is a strong argument for a preferential choice of this species as stabilizing component of ecosystems under climate change. Only recently, independently of each other, Zapater et al. (2011) and Jonard et al. (2011) found evidence for hydraulic lift of oak in young mixed beech-oak stands. This could be the cause for a lower sensitivity of beech to water scarcity in mixed beech-oak stands than in pure stands. Furthermore, Pretzsch et al. (2012) found that growth of beech in mixture with oak decreases less than in monoculture and thereby improves its competitive strength. This stress release by mixing as an important emergent property implying that the realized climatic niche of beech in mixed stands might be wider than in pure stands (Pretzsch et al. 2013). A further enrichment of mixed stands with typical ‘Quercetum’-species such as hornbeam (Carpinus betulus L.), lime (Tilia spp.) and Sorbus spp. or with drought tolerant species from present-day residual fragments of the postglacial northward migration (Quercus pubescens Willd., see Wellstein and Spada in press) or provenances (Thiel et al. 2012, Wellstein and Cianfaglione in review) can help to spread the risk further and reduce losses in production (Czajkowski and Bolte 2006, Bolte et al. 2009, Pretzsch and Schütze 2009). Modelling challenges Model comparisons are a powerful means to evaluate uncertainty in projections of ecosystem responses and are increasingly applied in global change studies (Pearson et al. 2006, Hlásny et al. 2011, Cheaib et al. 2012). SILVA and LandClim are two forest growth simulators developed for different purposes and correspondingly differ in underlying assumptions and structure (Pretzsch et al. 2008). SILVA was designed to project individual tree and stand development to assess growth and yield for practical management purposes. Major strength is its position-dependent competition calculation which allows for a precise reproduction of thinning and mixture practice alternatives. In contrast, LandClim was designed to simulate long-term forest dynamics under explicit consideration of competition, demographic processes, and disturbances (cf. Methods). The parametrization of 89
Manuscript 2: Climatic turning point for beech and oak LandClim emphasized the reproduction of succession dynamics and landcape-scale species distribution patterns. The two models’ respective strengths and weaknesses can be expected to highlight different aspects of uncertainty in estimates of future forest development and help to anticipate adverse consequences of climate change to facilitate reasonable adaptation measures (Littell et al. 2011). Despite their different structures both models gave very similar results and converged in their projection of the future development of beech-oak forests at the warm-dry distribution limit in Central Europe. Such agreement is rather atypical. Hlásny et al. (2011) for instance – after finding remarkable differences in the model output of BIOME-BGC and SYBILA for vegetation zones in Hungary – comment: ‘This appears to be general problem of such modelling […] The discrepancies […] emphasize the need for multi-model assessment of future forest development in order to clarify and highlight differences in model assumptions and model sensitivities to environmental and human intervention’. A possible reason for the congruence in the results is the calibration against the empirical data (cf. Methods; Fig. S1/2) as suggested by Mäkelä et al. (2000). Generally, process-based models incorporating tree physiology are given preference when addressing climate change issues (Mäkelä et al. 2000, Mäkelä 2003, Littell et al. 2011). They integrate physical site parameters (e.g. radiation, evapotranspiration, water availability) with basic physiological principles to calculate the photosynthetic potential (or more general carbon acquisition). Despite some freedom in species-specific parameter values, the physiological principles set reasonable limits to the possible growth even where parametrization data are scarce or contradictory. In contrast, models such as SILVA or LandClim employ empirical functions between growth parameter and growth reaction (cf. Bugmann 1994, Schumacher et al. 2004, Pretzsch 2009). Within their typical range, the direct relation between input (e.g. climate) and output variable (e.g. basal area growth) usually makes them much more precise than process-based models. Examples of approaches that promise to combine the strengths of either model type are hybrid models such as the BRIDGING model (Mäkelä 2003) or model coupling like the adjustment of SILVAs growth functions from BALANCE output (Moshammer et al. 2009). Since the first yield tables of the 18th and 19th century, forest growth models have come a long way. Today, forest management relies strongly on complex 90
Manuscript 2: Climatic turning point for beech and oak computer-aided growth simulators and expects decision support in face of a large uncertainty due to the projected climate change. However, it is clear that no single model solution can possibly cover the entire complexity of any ecosystem. Therefore, one of the most urgent future challenge for the scientific modelling community lies in overcoming structural differences in model in- and output (as done here) and thereby to ease multi-model inferences in the context of global change research based on suites of dynamic simulation models. Acknowledgements This study was funded by the Bavarian State Ministry of Sciences, Research and the Arts (StMWFK) within the framework of the ‘Bavarian Climate Programme 2020’ as part of the joint research centre ‘FORKAST’. The authors thank the German Meteorological Service (DWD) for providing data and the Bavarian State Ministry for Food, Agriculture and Forestry Project D03 for the scientific monitoring of strict forest reserves. References Allen, C. D., Macalady, A. K., Chenchouni, H., Bachelet, D., McDowell, N., Vennetier, M., Kitzberger, T., Rigling, A., Breshears, D. D., Hogg, E., Gonzalez, P., Fensham, R., Zhang, Z., Castro, J., Demidova, N., Lim, J.-H., Allard, G., Running, S. W., Semerci, A. and Cobb, N. 2010. A global overview of drought and heat-induced tree mortality reveals emerging climate change risks for forests. – Forest Ecology and Management 259: 660-684. Baltzer, J. L. and Thomas, S. C. 2007. Determinants of whole-plant light requirements in Bornean rain forest tree saplings. – Journal of Ecology 95: 1208-1221. Barriopedro, D., Fischer, E. M., Luterbacher, J., Trigo, R. M. and Garcia-Herrera, R. 2011. The hot summer of 2010: Redrawing the temperature record map of Europe. – Science 332: 220-224. Beaudet, M., Brisson, J., Gravel, D. and Messier, C. 2007. Effect of a major canopy disturbance on the coexistence of Acer saccharum and Fagus grandifolia in the understorey of an old-growth forest. – Journal of Ecology 95: 458-467. Bellassen, V., Le Maire, G., Guin, O., Dhôte, J., Ciais, P. and Viovy, N. 2011. Modelling forest management within a global vegetation model - Part 2: Model validation from a tree to a continental scale. – Ecological Modelling 222: 57-75. Bigler, C., Bräker, O. U., Bugmann, H., Dobbertin, M. and Rigling, A. 2006. Drought as an inciting mortality factor in scots pine stands of the Valais, Switzerland. – Ecosystems 9: 330- 343. Bolte, A., Ammer, C., Löf, M., Madsen, P., Nabuurs, G.-J., Schall, P., Spathelf, P. and Rock, J. 2009. Adaptive forest management in Central Europe: Climate change impacts, strategies and integrative concept. – Scandinavian Journal of Forest Research 24: 473-482. Bolte, A., Czajkowski, T. and Kompa, T. 2007. The north-eastern distribution range of European beech - A review. – Forestry 80: 413-429. 91
Manuscript 2: Climatic turning point for beech and oak Bonn, S. 2000. Competition dynamics in mixed beech-oak stands and its modifications expected due to climate changes. – Allgemeine Forst und Jagd Zeitung 171: 81-88. Boyden, S. B., Reich, P. B., Puettmann, K. J. and Baker, T. R. 2009. Effects of density and ontogeny on size and growth ranks of three competing tree species. – Journal of Ecology 97: 277- 288. Bréda, N., Huc, R., Granier, A. and Dreyer, E. 2006. Temperate forest trees and stands under severe drought: A review of ecophysiological responses, adaptation processes and long-term consequences. – Annals of Forest Science 63: 625-644. Briner, S., Elkin, C., Huber, R. and Grêt-Regamey, A. 2012. Assessing the impacts of economic and climate changes on land-use in mountain regions: A spatial dynamic modeling approach. – Agriculture, Ecosystems & Environment 149: 50 63.‐ Bugmann, H. 1994. On the ecology of mountainous forests in a changing climate: A simulation study. Ph.D. thesis no. 10638 – Swiss Federal Institute of Technology. Bugmann, H. and Cramer, W. 1998. Improving the behaviour of forest gap models along drought gradients. – Forest Ecology and Management 103: 247-263. Bugmann, H. K. M. and Solomon, A. M. 2000. Explaining forest composition and biomass across multiple biogeographical regions. – Ecological Applications 10: 95. Bunn, A. G. 2008. A dendrochronology program library in R (dplR). – Dendrochronologia 26: 115-124. Burkett, V. R., Wilcox, D. A., Stottlemyer, R., Barrow, W., Fagre, D., Baron, J., Price, J., Nielsen, J. L., Allen, C. D., Peterson, D. L., Ruggerone, G. and Doyle, T. 2005. Nonlinear dynamics in ecosystem response to climatic change: Case studies and policy implications. – Ecological Complexity 2: 357-394. Carnicer, J., Coll, M., Ninyerola, M., Pons, X., Sanchez, G. and Penuelas, J. 2011. Widespread crown condition decline, food web disruption, and amplified tree mortality with increased climate change-type drought. – Proceedings of the National Academy of Sciences 108: 1474- 1478. Cheaib, A., Badeau, V., Boe, J., Chuine, I., Delire, C., Dufrêne, E., François, C., Gritti, E. S., Legay, M., Pagé, C., Thuiller, W., Viovy, N. and Leadley, P. 2012. Climate change impacts on tree ranges: Model intercomparison facilitates understanding and quantification of uncertainty. – Ecology Letters 15: 533-544. Colombaroli, D., Henne, P. D., Kaltenrieder, P., Gobet, E. and Tinner, W. 2010. Species responses to fire, climate and human impact at tree line in the Alps as evidenced by palaeo-environmental records and a dynamic simulation model. – Journal of Ecology 98: 1346-1357. Cook, E. R. and Peters, K. 1997. Calculating unbiased tree-ring indices for the study of climatic and environmental change. – The Holocene 7: 361-370. Czajkowski, T. and Bolte, A. 2006. Unterschiedliche Reaktionen deutscher und polnischer Herkünfte der Buche (Fagus sylvatica L.) auf Trockenheit. – Allgemeine Forst und Jagd Zeitung: 30-40. Czúcz, B., Gálhidy, L. and Mátyás, C. 2011. Present and forecasted xeric climatic limits of beech and sessile oak distribution at low altitudes in Central Europe. – Annals of Forest Science 68: 99-108. Davi, H., Barbaroux, C., Francois, C. and Dufrêne, E. 2009. The fundamental role of reserves and hydraulic constraints in predicting LAI and carbon allocation in forests. – Agricultural and Forest Meteorology 149: 349-361. 92
Manuscript 2: Climatic turning point for beech and oak Elkin, C., Reineking, B., Bigler, C. and Bugmann, H. 2012. Do small-grain processes matter for landscape scale questions? Sensitivity of a forest landscape model to the formulation of tree growth rate. – Landscape Ecology 27: 697 711.‐ Ellenberg, H. 2009. Vegetation ecology of Central Europe. Translated by Gordon K. Strutt – Cambridge University Press. Fabbio, G., Bertini, G., Calderisi, M. and Ferretti, M. 2006. Status and trend of tree growth and mortality rate at the CONECOFOR plots, 1997-2004. – Annali dell'Istituto Sperimentale per la Selvicoltura: 11-19. Friedrichs, D. A., Trouet, V., Büntgen, U., Frank, D. C., Esper, J., Neuwirth, B. and Löffler, J. 2009. Species-specific climate sensitivity of tree growth in Central-West Germany. – Trees 23: 729-739. Fuhrer, J., Beniston, M., Fischlin, A., Frei, C., Goyette, S., Jasper, K. and Pfister, C. 2006. Climate risks and their impact on agriculture and forests in Switzerland. – Climatic Change 79: 79-102. García-Suárez, A., Butler, C. and Baillie, M. 2009. Climate signal in tree-ring chronologies in a temperate climate: A multi-species approach. – Dendrochronologia 27: 183-198. Gitlin, A. R., Sthultz, C. M., Bowker, M. A., Stumpf, A., Paxton, K. L., Kennedy, K., Muñoz, A., Bailey, J. K. and Whitham, T. G. 2006. Mortality gradients within and among dominant plant populations as barometers of ecosystem change during extreme drought. – Conservation Biology 20: 1477-1486. Gómez-Aparicio, L., García-Valdés, R., Ruiz-Benito, P. and Zavala, M. A. 2011. Disentangling the relative importance of climate, size and competition on tree growth in Iberian forests: implications for forest management under global change. – Global Change Biology 17: 2400-2414. Granier, A., Reichstein, M., Bréda, N., Janssens, I., Falge, E., Ciais, P., Grünwald, T., Aubinet, M., Berbigier, P., Bernhofer, C., Buchmann, N., Facini, O., Grassi, G., Heinesch, B., Ilvesniemi, H., Keronen, P., Knohl, A., Köstner, B., Lagergren, F., Lindroth, A., Longdoz, B., Loustau, D., Mateus, J., Montagnani, L., Nys, C., Moors, E., Papale, D., Peiffer, M., Pilegaard, K. and Pita, G., Pumpanen, J., Rambal, S., Rebmann, C., Rodrigues, A., Seufert, G., Tenhunen, J., Vesala, T., Wang, Q. 2007. Evidence for soil water control on carbon and water dynamics in European forests during the extremely dry year: 2003. – Agricultural and Forest Meteorology 143: 123- 145. Henne, P. D., Elkin, C. M., Reineking, B., Bugmann, H. and Tinner, W. 2011. Did soil development limit spruce (Picea abies) expansion in the Central Alps during the Holocene? Testing a palaeobotanical hypothesis with a dynamic landscape model. – Journal of Biogeography 38: 933-949. Henne, P. D., Elkin, C., Colombaroli, D., Samartin, S., Bugmann, H., Heiri, O. and Tinner, W. 2012. Impacts of changing climate and land use on vegetation dynamics in a Mediterranean ecosystem: Insights from paleoecology and dynamic modeling. – Landscape Ecology. Hlásny, T., Barcza, Z., Fabrika, M., Balázs, B., Churkina, G., Pajtík, J., Sedmák, R. and Turcáni, M. 2011. Climate change impacts on growth and carbon balance of forests in Central Europe. – Climate Research 47: 219-236. IPCC (ed.) 2012. Managing the risks of extreme events and disasters to advance climate change adaption. A special report of working groups I and II of the Intergovernmental Panel on Climate Change – Cambridge University Press. 93
Manuscript 2: Climatic turning point for beech and oak Johnstone, J. F., McIntire, E. J. B., Pedersen, E. J., King, G. and Pisaric, M. J. F. 2010. A sensitive slope: Estimating landscape patterns of forest resilience in a changing climate. – Ecosphere 1: art14. Jonard, F., André, F., Ponette, Q., Vincke, C. and Jonard, M. 2011. Sap flux density and stomatal conductance of European beech and common oak trees in pure and mixed stands during the summer drought of 2003. – Journal of Hydrology 409: 371-381. Jump, A. S., Hunt, J. M. and Peñuelas, J. 2006. Rapid climate change-related growth decline at the southern range edge of Fagus sylvatica. – Global Change Biology 12: 2163-2174. Köhl, M., Hildebrandt, R., Olschofksy, K., Köhler, R., Rötzer, T., Mette, T., Pretzsch, H., Köthke, M., Dieter, M., Abiy, M., Makeschin, F. and Kenter, B. 2010. Combating the effects of climatic change on forests by mitigation strategies. – Carbon Balance and Management 5: 8. Kölling, C. 2007. Klimahüllen für 27 Waldbaumarten. – AFZ-DerWald 23: 1242-1245. Kreienkamp, F., Spekat, A. and Enke, W. 2009. Sensitivity studies with a statistical downscaling method – The role of the driving large scale model. – Meteorologische Zeitschrift 18: 597-606. Lebourgeois, F., Bréda, N., Ulrich, E. and Granier, A. 2005. Climate-tree-growth relationships of European beech (Fagus sylvatica L.) in the French Permanent Plot Network (RENECOFOR). – Trees 19: 385-401. Leuschner, C., Backes, K., Hertel, D., Schipka, F., Schmitt, U., Terborg, O. and Runge, M. 2001. Drought responses at leaf, stem and fine root levels of competitive Fagus sylvatica L. and Quercus petraea (Matt.) Liebl. trees in dry and wet years. – Forest Ecology and Management 149: 33-46. Linder, M. 2000. Developing adaptive forest management strategies to cope with climate change. – Tree Physiology 20: 299-307. Lindner, M., Maroschek, M., Netherer, S., Kremer, A., Barbati, A., Garcia-Gonzalo, J., Seidl, R., Delzon, S., Corona, P., Kolström, M., Lexer, M. J. and Marchetti, M. 2010. Climate change impacts, adaptive capacity, and vulnerability of European forest ecosystems. – Forest Ecology and Management 259: 698-709. Littell, J. S., McKenzie, D., Kerns, B. K., Cushman, S. and Shaw, C. G. 2011. Managing uncertainty in climate-driven ecological models to inform adaptation to climate change. – Ecosphere 2: art102. Mäkelä, A. 2003. Process-based modelling of tree and stand growth: Towards a hierarchical treatment of multiscale processes. – Canadian Journal of Forest Research 33: 398-409. Mäkelä, A., Landsberg, J., Ek, A. R., Burk, T. E., Ter-Mikaelian, M., Agren, G. I., Oliver, C. D. and Puttonen, P. 2000. Process-based models for forest ecosystem management: Current state of the art and challenges for practical implementation. – Tree Physiology 20: 289-298. Manion, P. D. 1981. Tree disease concepts – Prentice-Hall. Mantgem, P. J. van and Stephenson, N. L. 2007. Apparent climatically induced increase of tree mortality rates in a temperate forest. – Ecology Letters 10: 909-916. McDowell, N. G., White, S. and Pockman, W. T. 2008a. Transpiration and stomatal conductance across a steep climate gradient in the southern Rocky Mountains. – Ecohydrology 1: 193-204. McDowell, N., Pockman, W. T., Allen, C. D., Breshears, D. D., Cobb, N., Kolb, T., Plaut, J., Sperry, J., West, A., Williams, D. G. and Yepez, E. A. 2008b. Mechanisms of plant survival and mortality during drought: Why do some plants survive while others succumb to drought? – New Phytologist 178: 719-739. 94
Manuscript 2: Climatic turning point for beech and oak Mérian, P. and Lebourgeois, F. 2011. Size-mediated climate–growth relationships in temperate forests: A multi-species analysis. – Forest Ecology and Management 261: 1382-1391. Michelot, A., Bréda, N., Damesin, C. and Dufrêne, E. 2012. Differing growth responses to climatic variations and soil water deficits of Fagus sylvatica, Quercus petraea and Pinus sylvestris in a temperate forest. – Forest Ecology and Management 265: 161-171. Moshammer, R., Rötzer, T. and Pretzsch, H. 2009. Analysing forest development under changing environmental conditions – Novel information for the forest management by combining models demonstrated for the forest enterprise Zittau. – Forest Ecology, Landscape Research and Nature Conservation 8: 63-70. Osada, N. 2012. Crown exposure to light and tree allometry of 11 tree species in a snowy cooltemperate forest in Japan. – Plant Ecology (formerly Vegetatio) 213: 783-794. Pearson, R. G., Thuiller, W., Araújo, M. B., Martinez-Meyer, E., Brotons, L., McClean, C., Miles, L., Segurado, P., Dawson, T. P. and Lees, D. C. 2006. Model-based uncertainty in species range prediction. – Journal of Biogeography 33: 1704-1711. Pérez-Ramos, I. M., Urbieta, I. R., Zavala, M. A. and Marañón, T. 2012. Ontogenetic conflicts and rank reversals in two Mediterranean oak species: Implications for coexistence. – Journal of Ecology 100: 467-477. Piedallu, C., Gégout, J.-C., Bruand, A. and Seynave, I. 2011. Mapping soil water holding capacity over large areas to predict potential production of forest stands. – Geoderma 160: 355-366. Piovesan, G., Biondi, F., di Filippo, A., Alessandrini, A. and Maugeri, M. 2008. Drought-driven growth reduction in old beech (Fagus sylvatica L.) forests of the central Apennines, Italy. – Global Change Biology 14: 1265-1281. Pollastrini, M., Desotgiu, R., Cascio, C., Bussotti, F., Cherubini, P., Saurer, M., Gerosa, G. and Marzuoli, R. 2010. Growth and physiological responses to ozone and mild drought stress of tree species with different ecological requirements. – Trees 24: 695-704. Pretzsch, H. 2009. Mixture proportion. – In: Pretzsch, H. (ed.), Forest dynamics, growth, and yield. Springer. Pretzsch, H. and Schütze, G. 2009. Transgressive overyielding in mixed compared with pure stands of Norway spruce and European beech in Central Europe: Evidence on stand level and explanation on individual tree level. – European Journal of Forest Research 128: 183-204. Pretzsch, H., Biber, P. and Ďurský, J. 2002. The single tree-based stand simulator SILVA: Construction, application and evaluation. – Forest Ecology and Management 162: 3-21. Pretzsch, H., Bielak, K., Block, J., Bruchwald, A., Dieler, J., Ehrhart, H.-P., Kohnle, U., Nagel, J., Spellmann, H., Zasada, M. and Zingg, A. 2013. Productivity of mixed versus pure stands of oak (Quercus petraea (Matt.) Liebl. and Quercus robur L.) and European beech (Fagus sylvatica L.) along an ecological gradient. – European Journal of Forest Research. Pretzsch, H., Dieler, J., Seifert, T. and Rötzer, T. 2012. Climate effects on productivity and resource-use efficiency of Norway spruce (Picea abies L. Karst.) and European beech (Fagus sylvatica L.) in stands with different spatial mixing patterns. – Trees 26: 1343-1360. Pretzsch, H., Grote, R., Reineking, B., Rötzer, T. and Seifert, S. 2008. Models for forest ecosystem management: A European perspective. – Annals of Botany 101: 1065-1087. R Development Core Team 2012. R: A language and environment for statistical computing, reference index version 2.15.2 – R Foundation for Statistical Computing – www.R-project.org. 2012-12-09. 95
Manuscript 2: Climatic turning point for beech and oak Raftoyannis, Y. and Radoglou, K. 2002. Physiological responses of beech and sessile oak in a natural mixed stand during a dry summer. – Annals of Botany 89: 723-730. Rehfeldt, G. E., Ferguson, D. E. and Crookston, N. L. 2009. Aspen, climate, and sudden decline in western USA. – Forest Ecology and Management 258: 2353-2364. Rinn, R. 2003. TSAP-Win (time series analysis and presentation for dendrochronology and related applications) Version 0.53 User Reference. Rötzer, T., Dieler, J., Mette, T., Moshammer, R. and Pretzsch, H. 2010. Productivity and carbon dynamics in managed Central European forests depending on site conditions and thinning regimes. – Forestry 83: 483-496. Sánchez-Gómez, D., Zavala, M. A., Schalkwijk, D. B., Urbieta, I. R. and Valladares, F. 2008. Rank reversals in tree growth along tree size, competition and climatic gradients for four forest canopy dominant species in Central Spain. – Annals of Forest Science 65: 605. Schär, C., Vidale, P. L., Lüthi, D., Frei, C., Häberli, C., Liniger, M. A. and Appenzeller, C. 2004. The role of increasing temperature variability in European summer heatwaves. – Nature 427: 332-336. Scharnweber, T., Manthey, M., Criegee, C., Bauwe, A., Schröder, C. and Wilmking, M. 2011. Drought matters – Declining precipitation influences growth of Fagus sylvatica L. and Quercus robur L. in north-eastern Germany. – Forest Ecology and Management 262: 947-961. Scherrer, D., Bader, M. K.-F. and Körner, C. 2011. Drought-sensitivity ranking of deciduous tree species based on thermal imaging of forest canopies. – Agricultural and Forest Meteorology 151: 1632-1640. Schroeder, P., Brown, S., Mo, J. M., Birdsey, R. and Cieszewski, C. 1997. Biomass estimation for temperate broadleaf forests of the United States using inventory data. – Forest Science 43: 424- 434. Schumacher, S. and Bugmann, H. 2006. The relative importance of climatic effects, wildfires and management for future forest landscape dynamics in the Swiss Alps. – Global Change Biology 12: 1435-1450. Schumacher, S., Bugmann, H. and Mladenoff, D. 2004. Improving the formulation of tree growth and succession in a spatially explicit landscape model. – Ecological Modelling 180: 175 194.‐ Schumacher, S., Reineking, B., Sibold, J. and Bugmann, H. 2006. Modeling the impact of climate and vegetation on fire regimes in mountain landscapes. – Landscape Ecology 21: 539-554. Soja, A. J., Tchebakova, N. M., French, N. H. F., Flannigan, M. D., Shugart, H. H., Stocks, B. J., Sukhinin, A. I., Parfenova, E. I., Chapin, F. S. and Stackhouse, J. P. W. 2007. Climate-induced boreal forest change: Predictions versus current observations. – Global and Planetary Change 56: 274-296. Solberg, S., Dobbertin, M., Reinds, G. J., Lange, H., Andreassen, K., Fernandez, P. G., Hildingsson, A. and Vries, W. de 2009. Analyses of the impact of changes in atmospheric deposition and climate on forest growth in European monitoring plots: A stand growth approach. – Forest Ecology and Management 258: 1735-1750. Spiecker, H., Hansen, J., Klimo, E., Skovsgaard, J. P., Sterba, H. and Teuffel, K. von 2004. Norway spruce conversion – Options and consequences. Sykes, M. and Prentice, C. I. 1996. Climate change, tree species distributions and forest dynamics: A case study in the mixed conifer/northern hardwoods zone of northern Europe. – Climatic Change 34. 96
Manuscript 2: Climatic turning point for beech and oak Szabados, I. 2006. The effect of the precipitation on the tree ring width. – Carpathian Journal of Earth and Environmental Science 1: 39-44. Thiel, D., Nagy, L., Beierkuhnlein, C., Huber, G., Jentsch, A., Konnert, M. and Kreyling, J. 2012. Uniform drought and warming responses in Pinus nigra provenances despite specific overall performances. – Forest Ecology and Management 270: 200-208. van der Werf, G., Sass-Klaassen, U. G. and Mohren, G. 2007. The impact of the 2003 summer drought on the intra-annual growth pattern of beech (Fagus sylvatica L.) and oak (Quercus robur L.) on a dry site in the Netherlands. – Dendrochronologia 25: 103-112. Walter, H. and Lieth, H. 1967. Klimadiagramm-Weltatlas – VEB Gustav Fischer Verlag. Wellstein, C. and Cianfaglione, C. in review. Impact of geographic origin, extreme drought and gradual warming on growth characteristics in juvenile Quercus pubescens Willd. – Folia Geobotanica. Wellstein, C. and Spada, F. in press. The status of Quercus pubescens Willd. in Europe. Williams, J. W. and Jackson, S. T. 2007. Novel climates, no-analog communities, and ecological surprises. – Frontiers in Ecology and the Environment 5: 475-482. Worrall, J. J., Egeland, L., Eager, T., Mask, R. A., Johnson, E. W., Kemp, P. A. and Shepperd, W. D. 2008. Rapid mortality of Populus tremuloides in south-western Colorado, USA. – Forest Ecology and Management 255: 686-696. Zang, C., Rothe, A., Weis, W. and Pretzsch, H. 2011. Tree suitability under climate change conditions: Susceptibility of major forest tree species from tree-rings widths. – Allgemeine Forst und Jagd Zeitung 182: 98-112. Zapater, M., Hossann, C., Bréda, N., Bréchet, C., Bonal, D. and Granier, A. 2011. Evidence of hydraulic lift in a young beech and oak mixed forest using 18O soil water labelling. – Trees 25: 885-894. 97
Manuscript 3: Symmetric and asymmetric competition Introduction The joint effect of climate, light, nutrients and biotic interactions determines forest growth at a particular site. Out of the large amount of biotic interactions competition is supposed to be among the most important factors (Sánchez-Gómez et al. 2008, Kunstler et al. 2011). Until now it is not conclusively resolved how site conditions and competition among trees act together (individuals or species level; cf. Freckleton and Watkinson 2001). To investigate the interaction of competition with environmental stress, communities in Mediterranean regions are particularly interesting because they cover a long aridity gradient reaching close to the abiotic limits of the studied species. Competition is an interaction between organisms or species that determines the share of resource allocation and in which the fitness of one organism is lowered by the presence of another. It can be split into a symmetric, size-relative and asymmetric component (Weiner 1990). Symmetric competition means that individuals share resources whereas asymmetric competition implies that one species is able to dominate the resource. Size-relative competition can be considered as a special case of asymmetric competition (or symmetric, depending on the point of view) where resource allocation is proportional to the size of the individual. Symmetric competition is assumed mainly for belowground resources such as water and nutrients (Weiner et al. 1997, Cahill and Casper 2000 but see also Leuschner et al. 2001). Asymmetric competition is assumed mainly for the resource light (Coomes and Grubb 2000). Both, symmetric and asymmetric competition decrease resource availability for an organism and subsequently increase stress. The tolerance of an organism to one stress, e.g. drought stress, is typically reduced by other co-occurring stress agents such as shading due to asymmetric competition (Hereford 2009). The shade-drought tolerance trade-off implies that there are morphological and physiological reasons that species or individuals are either shade or drought tolerant. Shade and drought tolerance can be defined as the ability to maintain growth in deeper shade or under fewer water supply. Under low light conditions, plants invest proportionally more biomass in leaves and aboveground parts. This increases the transpiration surface relative to the amount of roots and, consequently, the susceptibility to dry conditions (Smith and Huston 1989). Conflicting requirements for shade and drought tolerances have been sup- 104
Manuscript 3: Symmetric and asymmetric competition ported by several empirical studies (Kubiske et al. 1996, Valladares and Niinemets 2008, Hallik et al. 2009, Montgomery et al. 2010). Niinemets and Valladares (2006) observed significant negative correlations between shade and drought tolerance rankings, except for evergreen angiosperms. However, this research topic is apparently not conclusively resolved as also contrary patterns were found (Sack 2004, Martínez-Tillería et al. 2012). A reason for conflicting results might be a negative correlation between shading and drought stress due to lower temperatures and evapotranspiration under closed canopies (Sack and Grubb 2002). This negative correlation is considered the reason for a positive effect of shading resulting in facilitation, mainly during establishment. The shade-drought tolerance trade-off can be assessed at the intraspecific and interspecific level. At the interspecific level it can be ask if shade tolerant species are less drought tolerant than shade intolerant species (cf. Niinemets and Valladares 2006). At the intraspecific level it can be ask if trees grown under dry conditions and subsequently adapted to drought, are more sensitive to shading (Kubiske and Abrams 1994, Ruiz-Benito et al. 2013). The change of the effect of asymmetric competition on growth under increased aridity relates to the individual level. Another theory related to competition and changes of its effect along environmental gradients is known as the ‘supply and demand’ theory. In economy, the law of supply and demand is considered as one of the fundamental principles (cf. Thweatt 1983). It states that as supply increases the value of the resource (its price) will tend to drop or vice versa, and as demand increases the value will tend to increase or vice versa. This theory is widely used in economics and can be transferred into ecology. In an ecological context it states that resource availability of an individual is determined by absolute resource availability and the abundance of competitors at the same time (supply minus demand; Davis et al. 1998, Taylor et al. 1990). Unlike in economics in ecological systems supply is usually an external variable (e.g. amount of water) which does not feedback on demand (cf. Muth 1961). Equilibrium is therefore reached exclusively by changes in demand. If demand is less than supply the difference can be consumed by upcoming individuals, i.e. forest density can increase. If demand is higher than supply, individuals starve and subsequently die. Which individual will die in particular could be determined by chance, yet, it is more likely mediated by its ability to persist/avoid stress and to aggregate resources relative to its competitors. Along a resource gradient such as the aridity gradient, supply decreases. Thereby, the equi- 105
Manuscript 3: Symmetric and asymmetric competition librium stand density falls and competition intensity increases (Briones et al. 1998, Davis et al. 1998, Bennett et al. 2012). In the context of Grime´s C-S-R framework it has been stated that the importance of competition increases as disturbance and stress decline (cf. Grime 1977, Huston and Smith 1987), or comparably formulated that growth reductions due to competition relative to other factors falls with increasing abiotic stress (Brooker and Kikvidze 2008; be also aware of the ‘Tilman debate’ summed up in e.g. Goldberg et al. 1999, Brooker et al. 2005, Craine 2005). There are two possible mechanism behind this: The first one follows Liebig's law of the minimum, which states that only the most limiting resource is relevant, at least at the organism level (Odum and Barrett 2005). Under abiotic stress (such as drought) the organism is limited mainly by this stress and not by competitors. The second mechanism is a lower carrying capacity of high-stress environments (Tomé et al. 1996, Boisvenue and Running 2006) and thereby reduced stand basal area and subsequently less competitors. Variable climates like the Mediterranean produce fluctuating resource levels, which in turn might cause the size of populations (or stand density when referring to forests) to decrease below the level at which competition for resources occurs. Accordingly, it has been argued that competition might be less frequent in high-stress communities, though not less important (Fowler 1986, Vilà and Sardans 1999). The stress-gradient hypothesis agrees generally with Grime´s C-S-R framework but adds another aspect (Maestre et al. 2005, Lortie and Callaway 2006, Maestre et al. 2009, Pretzsch et al. 2013). It predicts that the frequency of facilitative and competitive interactions will vary inversely with increasing abiotic stress, with facilitation being more common when abiotic stress is high (Bertness and Callaway 1994, Holmgren et al. 1997, Maestre et al. 2006, Danger et al. 2008). It has been stated that under some conditions neighbours have positive effects that outbalance negative effects. Whereas neighbours lower soil water (belowground resource, symmetric competition) and reduce available light (aboveground resource, asymmetric competition) they might also have positive influences. Among the most important are modified micro-climate and soil conditions (Gómez-Aparicio et al. 2005, Aponte et al. 2011, Granda et al. 2012). A number of studies show that the nature of competition actually changes along ecological gradients. For grasslands Bennett et al. (2012) assessed the validity of 106
Manuscript 3: Symmetric and asymmetric competition Grime´s predictions and the patterns expected by the supply and demand theory. A decline in competition with productivity was found. For forests the mode of competition has been found to change from asymmetric to symmetric along a nutritional gradient from fertile to poor sites (Hara 1993, Pretzsch and Biber 2010). This indicates that under low nutrient supply the effect of competition for the target resource increases relative to the effect of competition for light. A similar pattern was observed by Zang et al. (2012) for Norway spruce. There was evidence for a change in competition mode from asymmetric competition under conditions with sufficient soil water supply to a more symmetric competition under dry conditions. An analysis of Kunstler et al. (2011) found that the importance of competition varied along the climatic gradient (degree-day sum and water availability) in a mountain forest of France. According to their analysis actual competition importance fell with increasing abiotic stress while intensity remains the same. Such changes in mode and effect of competition particularly along climatic gradients can even cause rank reversals in species performance (Sánchez-Gómez et al. 2008, Gómez-Aparicio et al. 2011, Scharnweber et al. 2011). Within this study we ask if the influence of symmetric and asymmetric competition on tree growth changes along the Mediterranean climatic gradient throughout the Iberian Peninsula. A statistical model for tree growth based on the National Forest Inventories of Spain was developed. The model considered the influence of climate, competition and tree size on individual tree growth. In order to formulate precise hypotheses it was necessary to distinguish between intensity and importance of competition (Brooker et al. 2005, Brooker and Kikvidze 2008, Freckleton et al. 2009, Kikvidze and Brooker 2010, Freckleton and Rees 2011). We defined competition intensity as growth reduction due to competition relative to growth in the absence of neighbours (but the same abiotic conditions). Given that competition has a negative effect on performance, intensity can be interpreted as the percentage of decrease in performance due to the presence of neighbours for given abiotic conditions, i.e. the absolute effect of competition is scaled to the local growth potential (Eq. 4). To competition importance we refer to as the negative effect of neighbours on growth relative to the impact of other (abiotic) factors (Eq. 5), i.e. the absolute effect of competition is scaled to the overall growth reduction. 107
Manuscript 3: Symmetric and asymmetric competition Using these indices we investigated the validity of three hypotheses motivated by the theories described above: 1. Intensity of asymmetric competition increases with aridity, according to the shade-drought tolerance trade-off. 2. Intensity of symmetric competition, e.g. competition for water, increases with aridity, according to the supply and demand theory. 3. Importance of symmetric and asymmetric competition compared to other factors falls with increasing aridity, according to Grime´s C-S-R framework. To address the first hypothesis the sensitivity to asymmetric competition under different abiotic conditions was compared. This could be done by comparison of the absolute effect of competition on growth (as difference between growth without neighbours and with neighbours). However, the change of this absolute difference along the gradient lacks its quantitative context and can therefore hardly be evaluated and compared among species. Intensity as percentage of growth reduction due to competition is scaled to the local growth potential. Similarly, the supply and demand theory was assessed by the change in intensity of symmetric competition. The intensity of competition as the percentage of growth reduction due to neighbours represents the local value of the target resource. The resource value is hypothesized to fall with supply, i.e. the intensity of competition falls. In contrast, the effect of competition on growth compared to other factors was assessed by the share of growth reduction by competition relative to the overall growth reduction (here only climate was considered). Methods Spanish National Forest Inventories Mediterranean regions occur in middle latitudes between parallels 30° to 40° North and South in five regions of the world: The Mediterranean Basin, California, central Chile, the Cape region in South-Africa and southwestern and southern Australia. Summer drought is the defining factor of the Mediterranean climate and particularly throughout the Iberian Peninsula, Spain (Köppen 1923). 108
Manuscript 3: Symmetric and asymmetric competition Annual rainfall ranges from about 1500 mm in montane forest to below 90 mm in the most arid sites. Annual mean temperatures range from about 11 °C to 17 °C. In the present study the Second and Third Spanish National Forest Inventories conducted between 1986-1996 and 1997–2007 were used (Villaescusa and Diaz 1998, Villanueva 2004). Survey plots were arranged in a regular 1 km raster. They consist of four concentric radii (i.e. the minimum tree diameter measured varied with the radius of the plot): All trees with diameter at breast height (DBH) of 7.5 cm were measured within a 5 m radius plot, trees with DBH = 12.5 cm within a 10 m radius plot, trees with DBH = 22.5 cm within a 15 m radius plot, and trees with DBH 42.5 cm within a 25 m radius plot. The 12 most frequent species were modelled (Pinus spp., Quercus spp. and Fagus sylvatica; refer to Tab. 2). Nonnative species were not modelled. Climatic data such as annual mean temperature (°C) and annual precipitation sum (mm) were provided by the Spanish Institute of Meteorology, as an interpolation of the information recorded in meteorological stations from 1971 to 2000 with 1 km spatial resolution (Gonzalo 2008). Bayesian non-linear model For this study we chose the annual ‘relative basal area increment’ (relBAI) as response variable (Eq. 1). Explaining variables were climate (temperature and precipitation), competition and tree size (Eq. 3). Relative basal area increment times hundred is the percentage of annual basal area increment for a given tree basal area. Thereby the pure geometric effect DBH on growth increment was eliminated. BA=π⋅( DBH 2) 2 relBAI=BA III −BA II BA II ⋅1 yrs Eq. 1 BA: Basal area (cm2). Index denotes second and third NFI. DBH: Diameter at breast height (cm) Numerous approaches exist to calculate measures for competition (Dale et al. 1985, von Oheimb et al. 2011). Two different competition measures were used within this study (non-spatially explicit, no differentiation of inter- and intraspecific competition): Stand basal area for symmetric competition SC and basal area of larger trees for asymmetric competition AC. To account for the inventory 109
Manuscript 3: Symmetric and asymmetric competition method, SC (for each plot) and AS (for each tree) were calculated under consideration of the reference area for each DBH class (Eq. 2). Within each survey plot the trees are ordered according to their DBH. SC p=100⋅ ∑ i=1 4∑ j=1 nπ 4⋅DBH2 Ai AC k=100⋅∑ i=1 4∑ j=k+1 nπ 4⋅DBH2 Ai Eq. 2 SCp: Symmetric competition in plot p as stand basal area in % of plot area. ACk: Asymmetric competition for target tree k as stand basal area of langer trees in % of plot area. DBHij: Diameter at breast height (cm) Ai: Plot area within DBH class of tree ij was measured (cm2) Iterator 4: Number of sub-plots with different radii in the Spanish National Forest Inventory. Symmetric competition was the same for all trees within a plot whereas asymmetric competition differed among individual trees. The stand basal area minus the basal area of the target tree is an alternative measure for symmetric competition. This measure differs among trees within the same plot (considered as neighbours) by their own size difference. The larger the target tree the smaller is the index compared to other trees within the same plot. Thereby an asymmetric component is included in this measure. Since we aimed at separating symmetric and asymmetric competition such an index would have been contra-productive. Mediterranean species distributions and tree growth show good correlations with annual mean temperature and annual precipitation sum (Thuiller et al. 2003, Sánchez-Gómez et al. 2008, Rodríguez-Sánchez et al. 2010, Gómez-Aparicio et al. 2011, Kunstler et al. 2011). Precipitation served as proxy for water availability and thereby aridity and drought stress. An interaction between temperature and precipitation was considered in the model. The relative basal area increment relBAI (Eq. 1) of individual trees was modelled as a non-linear function of climatic variables (temperature and precipitation), local interactions with neighbouring trees (symmetric and asymmetric competition) and tree diameter (DBH). All independent variables were normalized. Separate models were fitted for 12 selected species. 110
Manuscript 3: Symmetric and asymmetric competition We decided for the following model structure with a gamma error distribution as basis for backward model selection: relBAI=exp(α+ β T ⋅T+ β T2 ⋅T 2 + β P ⋅P+β P2 ⋅P 2 + β TP ⋅T⋅P+ β DBH ⋅DBH+ β DBH2 ⋅DBH 2 ) +β C ⋅(γ⋅SC+ (1−γ)⋅AS)+ β CT ⋅T⋅(γ T ⋅SC+ (1−γ T )⋅AS)+ β CP ⋅P⋅(γ P ⋅SC+ (1−γ P )⋅AS) + β CT2 ⋅T 2 ⋅(γ T2 ⋅SC+ (1−γ T )⋅AS)+ β CP2 ⋅P 2 ⋅(γ P2 ⋅SC+ (1−γ P )⋅AS)+ ϵ Eq. 3 T: Annual mean temperature (°C) P: Annual precipitation sum (mm) DBH: Diameter at breast height (cm) SC: Symmetric competition (% of basal area of neighbouring trees) AC: Asymmetric competition (% of basal area of taller trees) γ constrained to 0 ≤ γ ≤ 1; for specification of other priors and JAGS code see Supplement 1. For precipitation P, temperature T and DBH linear and quadratic terms were tested whereas for competition a linear relation was assumed. Additionally, an interaction for precipitation and temperature was included. Linear interaction terms for competition with climate allow for a change in absolute growth reduction due to competition along the gradient. Competition was divided into its symmetric and asymmetric component. Since they are naturally correlated (e.g. for the smallest tree in stand SC = AS; for the tallest AS = 0) they could not be used as two independent variables. Therefore, competition was included as a joint regression term. Within this term the effect (βC) of competition on relBAI was shared between SC and AS by parameters γ. Quadratic interaction terms for climate and competition allowed for a change of positive and negative interaction along the gradient (important when the species growth optimum is at intermediate values; Gómez-Aparicio et al. 2011). We used R.2.15.2 (R Development Core Team 2008) for data manipulations and JAGS 1.0.3 (Plummer 2003) for the Bayesian modelling (the R2jags package was used to interface between R and JAGS). We checked for convergence with two Monte Carlo Markov Chains using the potential scale reduction factor Rhat, setting our convergence threshold at Rhat < 1.01 and the Gelman and Gewke test for convergence (confidence = 0.95). We decided for a flexible burn-in period by resuming the mcmc after each 5,000 iterations if convergence was not reached, and a thinning of 2. After convergence was reached the two Monte Carlo Markov Chains were run another 10,000 iterations and the chains inspected visually. Model quality was evaluated additionally by calculation and visual inspection of the bias of prediction (0.9 < bias < 1.1) and residual diagnostics (QQ-plot inspec- 111
Manuscript 3: Symmetric and asymmetric competition tion). Due to high computation costs not the entire data was used to fit the model but a random selection of 10,000 trees (for each species). The final models were selected from the candidate models (Tab. 1) by the evaluation of DIC and R2. The null-model for the calculation of R2 constituted a model containing only intercept α. Furthermore, the principles of backward selection were applied, i.e. start with the full model (Eq. 3) and drop parameters on by one. Further decision rules to drop parameters: Quadratic terms before linear terms and interactions terms before non-interaction terms. Response curves for climate and competition were calculated for each species. Precipitation, temperature and symmetric and asymmetric competition were set to the dataset means, DBH to the species mean. Intensity and importance of competition Competition intensity measures growth reduction due to competition compared to abiotic factors. Changes in the intensity of competition along the precipitation and temperature gradient were analysed using the index C int =G −N −G +N max (G +N ;G −N ) Eq. 4, where G+N and G-N denote growth of the target species in the presence (+N) and absence (-N) of neighbours (Kikvidze and Brooker 2010, Brooker and Kikvidze 2008, Gaucherand et al. 2006). An intensity of 0.1 denotes a reduction of relative basal area increment by 10% due to interactions with neighbours. Competition importance is the proportional impact of competition relative to the overall impact 112 Tab. 1: Regression terms of the candidate models (cf. Eq. 3). Model βPβP2 βTβT2 βPT βDBH βDBH2 βCβCP βCT βCP2 βCT2 2 x x x x x x x x x x x 3 x x x x x x x x x x 4 xxxxxxxxxx 5 x x x x x x x x x 6 x x x x x x x x x 7 x x x x x x x x x 8 x x x x x x x x 9 x x x x x x x x
Manuscript 3: Symmetric and asymmetric competition of the environment. Index Cimp can be used to express changes in competition importance along productivity gradients (Eq. 5). C imp =G −N −G +N G max −min(G +N ;G −N ) Eq. 5 Again, G+N and G-N denote growth of the target species in the presence (+N) and absence (-N) of neighbours, Gmax is the growth without neighbours at the climatic optimum regarding precipitation and temperature. Since this index for competition importance depends on the abiotic growth optimum of each species it is highly sensitive to the abiotic range considered and has to be interpreted carefully. Competition intensity and importance were calculated along the precipitation and temperature gradient (but only shown for precipitation because temperature had almost no effect). For symmetric competition AC was set to zero and SC to the dataset mean, and vice versa. T was set to the 11 °C (data set mean was 11.5° C but P. uncinata did not occur at this temperature), DBH was set to the species means. The absolute increase or decrease of these two competition indices (Cint and Cimp) over the gradient depends on the definition of ‘with neighbours’. The more neighbours the stronger is the change of the indices along the climatic gradient (given that an interaction exists). If the interaction between the abiotic variable and competition is weak (and given that G+N < G-N) the indices could even increase for low competition values (low stand density or tall trees) and decrease for high values (high stand density or small trees). Stand basal area and thereby symmetric and asymmetric competition presumably change along the stress gradient, i.e. productivity decreases with increasing aridity (Tomé et al. 1996, Boisvenue and Running 2006). Under dry conditions individuals are subjected to lower absolute values of symmetric and asymmetric competition. Thereby, relevant levels of competition (values of SC and AC) decrease along the aridity gradient. ‘Actual’ SC and AC were calculated along the gradient using a linear regression. Competition intensity and importance were than calculated using these actual occurring values. These values were used to calculate the ‘actual’ competition intensity and importance (cf. Kunstler et al. 2011). 113
Manuscript 3: Symmetric and asymmetric competition Besides the modelled effect of climate on growth, a stronger effect of asymmetric competition on conifers than on broadleaved is in agreement with ecological knowledge, i.e. shade tolerances of species (Zavala et al. 2000, Niinemets and Valladares 2006; Tab. S3). Furthermore, the common pattern of lower plasticity in growth (Chambel et al. 2005, Valladares et al. 2007) of rather shadetolerant species (beech, oaks) than that of shade-intolerant species also appeared in our growth model (Sánchez-Gómez et al. 2006, Richter et al. 2012). However, it has to be considered that this plasticity was observed within a large spatial extent and might also be caused by genetic diversity due to local adaptation of provenances (for example for Pinus nigra; Jagielska et al. 2007, Lucic et al. 2010). Together with the response of species to climate induced aridity the studied species can be arranged in Grime´s C-S-R triangle: Broadleaved oaks and beech are more at the competitors side whereas coniferous pines are more at the stresstolerator side (Grime 1977). Besides direct effects of the independent variables on growth also interactions among them were included in the model structure. The interaction term for temperature and precipitation mostly had a positive sign. This supports the negative effect of aridity on growth (increased temperatures increase evapotranspiration which was attenuated by coinciding increase in precipitation). The interaction of competition and climate ranged in the same amplitude. Whereas the direct effect of competition was stronger for the asymmetric component the interaction with climate was stronger for the symmetric component (see also below ‘competition importance’). This is plausible because the resource water strongly changes with precipitation whereas the resource light changes only slightly due to the correlation of canopy density and aridity (or site capacity and aridity). Despite the present model gave reasonable results which are in agreement with former studies and ecological expectations, it had a relatively low fraction of explained variance R2. A hierarchical model accounting for variability among survey plots caused by other factors than climate and competition, insect infestations, soil fertility etc. could increase R2 for such a large dataset. However, we decided to only use measured variables since the aim of this study was not to make predictions on future tree growth at the same sites but to assess ecological principles. As long as non of the unknown variables which could be represented 120
Manuscript 3: Symmetric and asymmetric competition by the variable ‘survey site’ is correlated with the used ones general results will not differ. Shade-drought tolerance trade-off ‒ intensity of asymmetric competition At the interspecific level the shade-drought tolerance trade-off was observed for broadleaved angiosperms and coniferous gymnosperms (Niinemets and Valladares 2006). Conifers do not tolerate so well shading but are less vulnerable to drought due to strong stomata regulation to avoid hydraulic failures under water scarcity (‘stress avoiders’), whereas broadleaved suffer more from drought stress but are considered to be more shade tolerant (Bond 1989, Zavala et al. 2000, Hallik et al. 2009). The first hypothesis in the present study refers to an intraspecific shade-drought tolerance trade-off. We presumed that a shade-drought tolerance trade-off should lead to an increase in intensity of asymmetric competition with aridity. Interestingly, this pattern was only found for oaks. Relative to other tree species within the Mediterranean forest communities oaks are rather shade tolerant and drought sensitive angiosperms (Tab. S2). An explanation therefore might be the ability of oaks as broadleaved angiosperms to adjust leaf structure and photosynthesis related properties in response to shading and increased aridity (e.g. Niinemets 2010, Wyka et al. 2012 for shade; Abrams 1990 for drought). Coniferous pines have lower abilities to adjust their leave structure and therefore no such trade-off could be observed (Bond 1989). This finding leads to the expectation that broadleaved oaks and beech might loose their advantage of higher shade tolerance to some degree under increased aridity (as expected due to climate change). Supply and demand ‒ intensity of symmetric competition We hypothesized that the intensity of symmetric competition, e.g. competition for water, increases with aridity, according to the supply and demand theory. This proved to be true for all species. Assuming constant stand basal area a movement along the aridity gradient led to falling supply while demand remained unchanged. If demand remains unchanged and supply decreases, a shortage occurs, leading to a higher resource ‘price’, i.e. higher competition intensity. The relationship of symmetric competition intensity and stand basal area was similar to the relation of price and demand and can be interpreted as a movement on the demand curve in terms of microeconomics. In contrast, the relationship of symmetric competition 121
Manuscript 3: Symmetric and asymmetric competition intensity and supply is not similar to the relation of price and supply. This is due to the essential difference between most economic and ecological systems that there is no feedback of supply on changes in demand (mediated by price) (Rapport 1991). The economic principle of supply and demand (Thweatt 1983) can thus be found in Mediterranean forest communities, with the mentioned modification. Under more arid conditions like expected due to climate change this might lead to a more competitive environment regarding symmetric competition for water. Competition importance and aridity Changes of competition importance along the stress gradient are closely related to Grime´s C-S-R framework and the stress gradient hypothesis as well as to the Tillman debate. To address this topic it was essential to separate symmetric and asymmetric competition because they represent competition for different resources, here water and light. While water supply changes along the aridity gradient, light availability is not directly influenced. Contrary to expectations according to Grime´s C-S-R framework, the importance of competition rather increased with aridity (but refer to discussion on actual competition importance below). The strong effect of supply and demand in the studied communities provides a reason for the increase of symmetric competition importance. The effect of supply and demand was stronger than the general growth reduction due to aridity. This is in agreement with Tilman’s idea that competition moves from canopies to roots with increasing aridity (e.g Brooker et al. 2005). Also the shadedrought tolerance trade-off for oaks is in accordance with an increase of symmetric and asymmetric competition importance under aridity. However, for most pines at least asymmetric competition importance fell, i.e. growth reduction due to aridity actually gained importance (refer to Liebig's law of the minimum; Odum and Barrett 2005). This is in accordance with lack of a shade-drought tolerance trade-off for pines. According to the present findings in more arid environments competition might gain importance (Gómez-Aparicio et al. 2011). A more differentiated evaluation results in a shift of relative importance of symmetric and asymmetric competition in favour of the first (Zang et al. 2012). This is advantageous for pines and could outbalance negative effects of increased importance of overall competition in a probably more arid future. 122
Manuscript 3: Symmetric and asymmetric competition Actual competition intensity and importance Under dry conditions individuals actually are subjected to lower absolute values of symmetric and asymmetric competition. Accounting for changed stand basal area ‘actual’ competition intensity and importance were calculated. According to Kunstler et al. (2011) the actual importance of competition falls with increasing abiotic stress which is largely in accordance with the present study. However, the patterns for intensity differ among the studies. Contrary to Kunstler et al. (2011), particularly actual intensity of asymmetric competition changed with aridity. Differences in findings could not only be attributed to differences in the representation of competition and model structure but also to the climatic gradients. Whereas Kunstler et al. (2011) assessed a spatially smaller and generally moisture site in France the aridity gradient of the Spanish peninsula is longer as well as more extreme. Moreover, when actual competition intensity and importance are used it has to be considered that there is a reason for lower stand basal area: Carrying capacity of the site is lower due to lower resource supply. The regulation of stand basal area is at least partly mediated by competition and not only by abiotic growth reductions and increased mortality (Ruiz-Benito et al. 2013). Symmetric competition intensity as local value of the resource water was found to increase with aridity whereas the effect was smaller under consideration of actual occurring competition levels. This supports the presumption that symmetric competition is a candidate for regulation of stand basal area with decreasing resource supply, particularly for oaks. In contrast, actual asymmetric competition intensity strongly fell and is therefore not good a candidate. In the present study a consideration of actual levels of competition along the aridity gradient did result in a decrease of actual competition importance. This is in accordance with Grime´s C-S-R framework that is perhaps better addressed using actual occurring conditions (e.g. a consideration of actual occurring levels of competition). Particularly, the decrease in actual occurring asymmetric competition results in a strong loss of competitiveness of shade tolerant oaks with increasing aridity. Thereby, an underestimation of changes in competitiveness with aridity is likely when only the mere interplay of shade and drought tolerance with aridity is evaluated but changes in stand basal area are not considered. 123
Manuscript 3: Symmetric and asymmetric competition Conclusions For the Iberian Peninsula under climate change a temperature rise is expected while precipitation is projected to decrease. According to the IPCC ‘there is medium confidence that droughts will intensify in the 21st century in some seasons and areas, due to reduced precipitation and/or increased evapotranspiration’ (IPCC 2012, p. 14). Although Mediterranean species are adapted to summer drought, an impact on forests is expected because already today the conditions in drier regions are near their limit. In Mediterranean forests competition for both light and water strongly influences tree growth along the entire aridity gradient. The importance of competition even increases with aridity, with some exceptions. This indicates more competitive environments with proceeding climate change. The model indicated that particularly the effect of symmetric competition for water changes along the aridity gradient. Thereby growth reductions due to symmetric competition for water increase stronger than growth reductions due to asymmetric competition for light. This leads to the expectation that symmetric competition might gain importance in shaping Mediterranean forest communities under increased aridity. Shade tolerant and drought sensitive species such as broadleaved oaks and beech might thereby have a relative disadvantage under future climate, additionally to mere abiotic causes. This disadvantage might even be larger when changes of basal area due to site productivity and falling levels of actual asymmetric competition are considered. In any case, expected increase in aridity due to climate change has negative effects on all studied species. Acknowledgements This study was funded by the Bavarian State Ministry of Sciences, Research and the Arts (StMWFK) within the framework of the ‘Bavarian Climate Programme 2020’ in the joint research centre ‘FORKAST’. Further funding was obtained from the German Academic Exchange Service (DAAD), firstly, to participate in the 12th EEF Congress (Ávila) which helped to initiate the project, and secondly, to facilitate a research stay in which the project was further elaborated. 124
Manuscript 3: Symmetric and asymmetric competition References Abrams, M. D. 1990. Adaptations and responses to drought in Quercus species of North-America. – Tree Physiology 7: 227-238. Aponte, C., García, L. V., Pérez-Ramos, I. M., Gutiérrez, E. and Marañón, T. 2011. Oak trees and soil interactions in Mediterranean forests: A positive feedback model. – Journal of Vegetation Science 22: 856-867. Bennett, J. A., Cahill, J. F. and Chen, H. Y. H. 2012. Evaluating the relationship between competition and productivity within a native grassland. – PLoS ONE 7: e43703. Bertness, M. D. and Callaway, R. M. 1994. Positive interactions in communities. – Trends in Ecology & Evolution 9: 191-193. Boisvenue, C. and Running, S. W. 2006. Impacts of climate change on natural forest productivity - Evidence since the middle of the 20th century. – Global Change Biology 12: 862-882. Bond, W. J. 1989. The tortoise and the hare: Ecology of angiosperm dominance and gymnosperm persistence. – Biological Journal of the Linnean Society 36: 227-249. Briones, O., Montaña, C. and Ezcurra, E. 1998. Competition intensity as a function of resource availability in a semiarid ecosystem. – Oecologia 116: 365-372. Brooker, R. W. and Kikvidze, Z. 2008. Importance: An overlooked concept in plant interaction research. – Journal of Ecology 96: 703-708. Brooker, R. W., Kikvidze, Z., Pugnaire, F. I., Callaway, R. M., Choler, P., Lortie, C. J. and Michalet, R. 2005. The importance of importance. – Oikos 109: 63-70. Cahill, J. F. and Casper, B. B. 2000. Investigating the relationship between neighbor root biomass and belowground competition: Field evidence for symmetric competition belowground. – Oikos 90: 311-320. Chambel, M. R., Climent, J., Alía, R. and Valladares, F. 2005. Phenotypic plasticity: A useful framework for understanding adaptation in forest species. – Investigación Agraria: Sistemas y recursos forestales 14: 334-344. Coomes, D. A. and Grubb, P. J. 2000. Impact of root competition in forests and woodlands: A theoretical framework and review of experiments. – Ecological Monographs 70: 171-207. Craine, J. M. 2005. Reconciling plant strategy theories of Grime and Tilman. – Journal of Ecology 93: 1041-1052. Dale, V., Doyle, T. and Shugart, H. H. 1985. A comparison of tree growth models. – Ecological Modelling 29: 145-169. Danger, M., Daufresne, T., Lucas, F., Pissard, S. and Lacroix, G. 2008. Does Liebig's law of the minimum scale up from species to communities? – Oikos 117: 1741-1751. Davis, M. A., Wrage, K. J. and Reich, P. B. 1998. Competition between tree seedlings and herbaceous vegetation: Support for a theory of resource supply and demand. – Journal of Ecology 86: 652-661. Fowler, N. 1986. The role of competition in plant communities in arid and semiarid regions. – Annual Review of Ecology and Systematics 17: 89-110. Freckleton, R. P. and Rees, M. 2011. Problems formalising the concept of importance in ecology. – Trends in Ecology & Evolution 26: 498-499. Freckleton, R. P. and Watkinson, A. R. 2001. Asymmetric competition between plant species. – Functional Ecology 15: 615-623. 125
Manuscript 3: Symmetric and asymmetric competition Freckleton, R. P., Watkinson, A. R. and Rees, M. 2009. Measuring the importance of competition in plant communities. – Journal of Ecology 97: 379-384. Gaucherand, S., Liancourt, P. and Lavorel, S. 2006. Importance and intensity of competition along a fertility gradient and across species. – Journal of Vegetation Science 17: 455-464. Goldberg, D. E., Rajaniemi, T., Gurevitch, J. and Stewart-Oaten, A. 1999. Empirical approaches to quantifying interaction intensity: Competition and facilitation along productivity gradients. – Ecology 80: 1118-1131. Gómez-Aparicio, L., García-Valdés, R., Ruiz-Benito, P. and Zavala, M. A. 2011. Disentangling the relative importance of climate, size and competition on tree growth in Iberian forests: Implications for forest management under global change. – Global Change Biology 17: 2400-2414. Gómez-Aparicio, L., Gómez, J. M., Zamora, R. and Boettinger, J. L. 2005. Canopy vs. soil effects of shrubs facilitating tree seedlings in Mediterranean montane ecosystems. – Journal of Vegetation Science 16: 191-198. Gonzalo, J. 2008. Diagnosis fitoclimática de la España peninsular. Actualización y análisis geoestadístico aplicado – Universidad Politécnica de Madrid. Escuela Técnica Superior de Ingenieros de Montes. Granda, E., Escudero, A., La Cruz, M. de, Valladares, F. and Pillar, V. 2012. Juvenile-adult tree associations in a continental Mediterranean ecosystem: No evidence for sustained and general facilitation at increased aridity. – Journal of Vegetation Science 23: 164-175. Grime, J. P. 1977. Evidence for the existence of three primary strategies in plants and its relevance to ecological and evolutionary theory. – The American Naturalist 111: 1169. Hallik, L., Niinemets, Ü. and Wright, I. J. 2009. Are species shade and drought tolerance reflected in leaf-level structural and functional differentiation in northern hemisphere temperate woody flora? – New Phytologist 184: 257-274. Hara, T. 1993. Mode of competition and size-structure dynamics in plant communities. – Plant Species Biology 8: 75-84. Hereford, J. 2009. A quantitative survey of local adaptation and fitness trade offs. – The American ‐ Naturalist 173: 579-588. Holmgren, M., Scheffer, M. and Huston, M. A. 1997. The interplay of facilitation and competition in plant communities. – Ecology 78: 1966-1975. Huston, M. and Smith, T. 1987. Plant succession: Life history and competition. – The American Naturalist 130: 168. IPCC (ed.) 2012. Managing the risks of extreme events and disasters to advance climate change adaption. A special report of working groups I and II of the Intergovernmental Panel on Climate Change – Cambridge University Press. Jagielska, A., Cwalina, M. and Prus-Glowacki, W. 2007. Genetic diversity of black pine stands (Pinus nigra Arn.). – Sylwan 151: 23-31. Kikvidze, Z. and Brooker, R. W. 2010. Towards a more exact definition of the importance of competition - A reply to Freckleton et al. (2009). – Journal of Ecology 98: 719-724. Köppen, W. 1923. Die Klimate der Erde – de Gruyter. Kubiske, M. E. and Abrams, M. D. 1994. Ecophysiological analysis of woody species in contrasting temperate communities during wet and dry years. – Oecologia 98: 303-312. Kubiske, M. E., Abrams, M. D. and Mostoller, S. A. 1996. Stomatal and nonstomatal limitations of photosynthesis in relation to the drought and shade tolerance of tree species in open and understory environments. – Trees 11: 76. 126
Manuscript 3: Symmetric and asymmetric competition Kunstler, G., Albert, C. H., Courbaud, B., Lavergne, S., Thuiller, W., Vieilledent, G., Zimmermann, N. E. and Coomes, D. A. 2011. Effects of competition on tree radial-growth vary in importance but not in intensity along climatic gradients. – Journal of Ecology 99: 300-312. Leuschner, C., Hertel, D., Coners, H. and Büttner, V. 2001. Root competition between beech and oak: A hypothesis. – Oecologia 126: 276-284. Lortie, C. J. and Callaway, R. M. 2006. Re-analysis of meta-analysis: Support for the stressgradient hypothesis. – Journal of Ecology 94: 7-16. Lucic, A., Mladenovic-Drinic, S., Stavretovic, N., Isajev, V., Lavadinovic, V., Rakonjac, L. and Novakovic, M. 2010. Genetic diversity of Austrian pine (Pinus nigra Arn.) populations in Serbia revealed by RAPD. – Archives of Biological Sciences 62: 329-336. Maestre, F. T., Callaway, R. M., Valladares, F. and Lortie, C. J. 2009. Refining the stress-gradient hypothesis for competition and facilitation in plant communities. – Journal of Ecology 97: 199- 205. Maestre, F. T., Valladares, F. and Reynolds, J. F. 2005. Is the change of plant-plant interactions with abiotic stress predictable? A meta-analysis of field results in arid environments. – Journal of Ecology 93: 748-757. Maestre, F. T., Valladares, F. and Reynolds, J. F. 2006. The stress-gradient hypothesis does not fit all relationships between plant-plant interactions and abiotic stress: Further insights from arid environments. – Journal of Ecology 94: 17-22. Martínez-Tillería, K., Loayza, A. P., Sandquist, D. R., Squeo, F. A. and Ward, D. 2012. No evidence of a trade-off between drought and shade tolerance in seedlings of six coastal desert shrub species in north-central Chile. – Journal of Vegetation Science 23: 1051-1061. Montgomery, R. A., Reich, P. B. and Palik, B. J. 2010. Untangling positive and negative biotic interactions: Views from above and below ground in a forest ecosystem. – Ecology 91: 3641- 3655. Muth, J. F. 1961. Rational expectations and the theory of price movements. – Econometrica 29: 315-335. Niinemets, Ü. 2010. A review of light interception in plant stands from leaf to canopy in different plant functional types and in species with varying shade tolerance. – Ecological Research 25: 693-714. Niinemets, Ü. and Valladares, F. 2006. Tolerance to shade, drought, and water logging of temperate northern hemisphere trees and shrubs. – Ecological Monographs 76: 521-547. Odum, E. P. and Barrett, G. W. 2005. Fundamentals of ecology – Thomson Brooks/Cole. Oheimb, G. von, Lang, A. C., Bruelheide, H., Forrester, D. I., Wäsche, I., Yu, M. and Härdtle, W. 2011. Individual-tree radial growth in a subtropical broad-leaved forest: The role of local neighbourhood competition. – Forest Ecology and Management 261: 499-507. Papport, D. J. 1991. Myths in the foundations of economics and ecology. – Biological Journal of the Linnean Society 44: 185-202. Plummer, M. 2003. JAGS: A program for analysis of Bayesian graphical models using Gibbs sampling. – Proceedings of the 3rd International Workshop on Technische Universitat Wien Pretzsch, H. and Biber, P. 2010. Size-symmetric versus size-asymmetric competition and growth partitioning among trees in forest stands along an ecological gradient in Central Europe. – Canadian Journal of Forest Research 40: 370-384. 127
Manuscript 3: Symmetric and asymmetric competition Pretzsch, H., Bielak, K., Block, J., Bruchwald, A., Dieler, J., Ehrhart, H.-P., Kohnle, U., Nagel, J., Spellmann, H., Zasada, M. and Zingg, A. 2013. Productivity of mixed versus pure stands of oak (Quercus petraea (Matt.) Liebl. and Quercus robur L.) and European beech (Fagus sylvatica L.) along an ecological gradient. – European Journal of Forest Research. Rapport, D. J. 1991. Myths in the foundations of economics and ecology. – Biological Journal of the Linnean Society 44: 185-202. Richter, S., Kipfer, T., Wohlgemuth, T., Calderón Guerrero, C., Ghazoul, J. and Moser, B. 2012. Phenotypic plasticity facilitates resistance to climate change in a highly variable environment. – Oecologia 169: 269-279. Rodríguez-Sánchez, F., Hampe, A., Jordano, P. and Arroyo, J. 2010. Past tree range dynamics in the Iberian Peninsula inferred through phylogeography and palaeodistribution modelling: A review. – Review of Palaeobotany and Palynology 162: 507-521. Ruiz-Benito, P., Lines, E. R., Gómez-Aparicio, L., Zavala, M. A., Coomes, D. A. and Hector, A. 2013. Patterns and drivers of tree mortality in Iberian forests: Climatic effects are modified by competition. – PLoS ONE 8: e56843. Sack, L. 2004. Responses of temperate woody seedlings to shade and drought: do trade-offs limit potential niche differentiation? – Oikos 107: 110-127. Sack, L. and Grubb, P. 2002. The combined impacts of deep shade and drought on the growth and biomass allocation of shade-tolerant woody seedlings. – Oecologia 131: 175-185. Sánchez-Gómez, D., Valladares, F. and Zavala, M. A. 2006. Functional traits and plasticity in response to light in seedlings of four Iberian forest tree species. – Tree Physiology 26: 1425- 1433. Sánchez-Gómez, D., Zavala, M. A., Schalkwijk, D. B., Urbieta, I. R. and Valladares, F. 2008. Rank reversals in tree growth along tree size, competition and climatic gradients for four forest canopy dominant species in central Spain. – Annals of Forest Science 65: 605. Scharnweber, T., Manthey, M., Criegee, C., Bauwe, A., Schröder, C. and Wilmking, M. 2011. Drought matters – Declining precipitation influences growth of Fagus sylvatica L. and Quercus robur L. in north-eastern Germany. – Forest Ecology and Management 262: 947-961. Smith, T. and Huston, M. 1989. A theory of the spatial and temporal dynamics of plant communities. – Vegetatio 83: 49-69. Taylor, D. R., Aarssen, L. W. and Loehle, C. 1990. On the relationship between r/K selection and environmental carrying capacity: A new habitat templet for plant life history strategies. – Oikos 58: 239. Thuiller, W., Vayreda, J., Pino, J., Sabate, S., Lavorel, S. and Gracia, C. 2003. Large-scale environmental correlates of forest tree distributions in Catalonia (NE Spain). – Global Ecology and Biogeography 12: 313-325. Thweatt, W. O. 1983. Origins of the terminology ‘supply and demand’. – Scottish Journal of Political Economy 30: 287-294. Tomé, M., Ribeiro, F., Páscoa, F., Silva, R., Tavares, M., Palma, A. and Paulo, M. J. C. 1996. Growth trends in Portuguese forests: An exploratory analysis. – In: Spiecker, H. et al. (eds.), Growth trends in European forests. Springer Berlin Heidelberg, pp. 329–353. Valladares, F. and Niinemets, Ü. 2008. Shade tolerance, a key plant feature of complex nature and consequences. – Annual Review of Ecology, Evolution, and Systematics 39: 237-257. Valladares, F., Gianoli, E. and Gómez, J. M. 2007. Ecological limits to plant phenotypic plasticity. – New Phytologist 176: 749-763. 128
Manuscript 3: Symmetric and asymmetric competition Vilà, M. and Sardans, J. 1999. Plant competition in Mediterranean-type vegetation. – Journal of Vegetation Science 10: 281-294. Villaescusa, R. and Diaz, R. 1998. Segundo Inventario Forestal Nacional (1986–1996). Villanueva, J. A. 2004. Tercer Inventario Forestal Nacional (1997–2007). Weiner, J. 1990. Asymmetric competition in plant populations. – Trends in Ecology & Evolution 5: 360-364. Weiner, J., Wright, D. B. and Castro, S. 1997. Symmetry of below-ground competition between Kochia scoparia individuals. – Oikos 79: 85. Wyka, T. P., Oleksyn, J., Żytkowiak, R., Karolewski, P., Jagodziński, A. M. and Reich, P. B. 2012. Responses of leaf structure and photosynthetic properties to intra-canopy light gradients: A common garden test with four broadleaf deciduous angiosperm and seven evergreen conifer tree species. – Oecologia 170: 11-24. Zang, C., Pretzsch, H. and Rothe, A. 2012. Size-dependent responses to summer drought in Scots pine, Norway spruce and common oak. – Trees 26: 557-569. Zavala, M. A., Espelta, J. M. and Retana, J. 2000. Constraints and trade-offs in Mediterranean plant communities: The case of holm oak-Aleppo pine forests. – The Botanical Review 66: 119-149. Supplement Fig. S1: ‚Actual‛ competition intensity: Intensity of symmetric competition increased for most species or remained more or less unchanged. Asymmetric competition intensity fell for pines and some oaks. 129
Manuscript 4: Ecotype mixing as climate change adaptation tainty of the (long-term) performance of species or specific ecotypes (Donaldson- Matasci et al. 2008). The complexity of effects and possible advantages of mixed stands is currently intensively studied (Pretzsch et al. 2010, Pretzsch et al. 2012). However, there is an important difference between forest management and other fields of product management. In forests, self-thinning takes place due to competition among individuals. Forest management modifies this selection process further (Huuskonen and Hynynen 2006, Bradford and Palik 2009, Tikkanen et al. 2012). This natural process reduces the number of individuals considerably over time (Pickard 1983, Lonsdale 1990). Thereby, in forests, not all assets (i.e. individual trees) of the portfolio (the stand) remain until the date of harvest (i.e. they die due to competition). While in economic product management a drop out of single assets of the portfolio usually comes along with profit losses, in forests the drop out of tree individuals is necessary to provide space for the surviving trees to grow well. In the present study, two different plantation strategies appear apart from betting on the apparently best ecotype. The first strategy is to grow ecotypes in several single-ecotype stands, the second is to mix ecotypes within one stand (termed ‘ecotype mixing’). The type of planting strategy determines whether only individuals of the same or of several ecotypes compete. It can be expected that considering self-thinning together with ecotype mixing gives different results regarding yield depending on the planting strategy. We consider the competitive ability among tree individuals to be correlated with growth and general performance as yield (Bigler and Bugmann 2003, Shifley et al. 2006, Wunder et al. 2008, Luo and Chen 2011). In order to evaluate the combined effect of statistical averaging and self-thinning with ecotype mixing, we apply a simulation model which we term ‘maximum selection’. Compensatory effects, such as increased growth of tree individuals due to a release from competition as suggested in the insurance hypothesis, are not considered in this study. Two alternative models in addition to maximum selection are tested. Hereby lottery sampling is applied as a null-model, including only the effect of ecotype mixing but without a selection mechanism. Ecologically, this means that performance and survival during self-thinning and yield are uncorrelated. By means of this selection method the simulation model can further be related to the mere portfolio effect. Minimum selection is applied as the inversed scenario compared to maximum selection. Ecologically, this means that performance and survival during self-thinning and yield are negatively correlated. 136
Manuscript 4: Ecotype mixing as climate change adaptation In this study we utilize experimental data from common garden experiments (Kreyling et al. 2012A, Thiel et al. 2012) with important European tree species (Fagus sylvatica L. and Pinus nigra Arn.) as well as artificial data. Due to experimental constraints, the present study is based on sapling growth. Sapling growth during competition in the self-thinning phase might not adequately reflect the performance of adult trees. Nevertheless, we do not expect a negative correlation between growth at different ontogenetic stages if environmental conditions remain similar during the overall tree life span (or rotation period). By means of artificial data, constituting a known ‘truth’, mathematical artefacts in methodology can be detected (Austin 1976, Steinbauer et al. 2012). Furthermore, such analysis enables ecologists to test hypotheses in the lack of data. Thus, the simulations were scaled by the experimental data, whereby the artificial data aided in the understanding of the influence of the data structure. Expected yield was simulated for each dataset under different mixing strategies (no mixing and mixing of two to n ecotypes) and selection mechanisms (maximum selection, lottery sampling, minimum selection). This simulation model explores the effect of ecotype mixing, self-thinning, and statistical averaging on expected yield; the risk of low and the chance for high yield. The consequences of an increase of within-stand variability in biomass production by ecotype mixing on expected yields and their uncertainty are assessed. We hypothesize that the reduced risk of low yields due to the portfolio effect are not accompanied by lower chances for high yields in forest stands as a result of self-thinning. If this is the case, ecotype mixing constitutes an opportunity for the adaptation of forest stands to current climate change. Methods Experimental data Here, we focus on two species: Pinus nigra (European black pine), a generally drought-resistant species with a patchy (sub-) Mediterranean distribution range (Isajev 2003), which is a potential candidate to replace climate change-threatened conifer species such as Picea abies (Norway spruce) in Central Europe. We also consider Fagus sylvatica (European beech), which is the dominant native forest species in Central Europe yet considered threatened in the face of climate change (Geßler et al. 2006, Ohlemuller et al. 2006). The data used in this study stemmed from a common garden experiment (EVENT 3) in the Ecological-Botanical gar- 137
Manuscript 4: Ecotype mixing as climate change adaptation dens of the University of Bayreuth, Germany (Jentsch and Beierkuhnlein 2010, Beierkuhnlein et al. 2011) where the differences in performance among various ecotypes of key European forest species were studied. Seedlings of eight ecotypes of P. nigra (Thiel et al. 2012) and seven ecotypes of F. sylvatica (Kreyling et al. 2012a) were planted in pots and through 2009/2010 (P. nigra) and 2010 (F. sylvatica). The P. nigra dataset included 42 replicates per ecotype, whereas the F. sylvatica dataset comprised of 18 replicates per ecotype. The individuals of both species were irrigated according to the local daily 30-year average precipitation. Application occurred twice a week, using collected rain water. The height of the plants was recorded before and at the end of the growing season. The relative increment rate was determined as the difference between these two measurements in percent (Fig. 1 and 2). Artificial data In addition to the experimental data, artificial datasets with the same structure were created using R 2.15.2 (R Development Core Team 2012, R-Code provided in online Supplement 1). The artificial datasets consisted of seven ecotypes characterized by mean growth and variation (normal distributed data represented by 100 individuals). Means of the ecotypes were randomly selected from a normal distribution. Variation within each ecotype was held constant (Fig. 1A). This basic dataset A was modified to assess the influence of data structure on expected yields. Firstly, the means of the ecotypes were multiplied with a constant in order to increase differences between them (Fig. 1B, constant is 10). Next, the basic dataset was modified by creating an over-performing ecotype with significantly higher increment rates than others (Fig. 3C, adding 10 to ecotype g). The following data-scenario was the same as in Fig. 1C, i.e. one over-performer, but also contained an under-performing ecotype (Fig. 1D). The same was done with two over-performers (Fig. 1E), where one data scenario again contained an underperformer (Fig. 1F). The three different selection procedures described below were applied to these datasets in the same way as for the experimental data. Maximum selection Maximum selection mimicked an experiment where several individuals were hypothetically planted together. The best performing individual (i.e. best growing) of each plantation was assumed to survive and its growth was measured. The 138
Manuscript 4: Ecotype mixing as climate change adaptation underlying assumption was that performance/growth is closely linked to competitiveness and survival and thereby controls the result of self-thinning (Bigler and Bugmann 2003, Wunder et al. 2008). In the sampling procedure carried out in R 2.15.2 (R Development Core Team 2012, R-Code provided in online Supplement 2) the first step was to create a data subset for each possible combination of ecotypes. The resulting number of subsets corresponded to the binomial coefficient. Then, five individuals were selected randomly from a subset and the value of the best performing individual was saved. The number of five individuals determined the difference between mean performance of the ecotype and the increase due to maximum selection. As the same value for that parameter was used in all simulations, a change only resulted in absolute differences but not relative differences between resulting patterns. Ecologically, this parameter represents the number of individuals which compete in a stand in first order neighbourhood, for which five seems to be a reasonable estimate. This selection rule was repeated 5000 times for each subset. For each subset, the minimum, median, maximum and the quantiles 0.05, 0.25, 0.75, and 0.95 were calculated from the repetitions. Quantile data was then grouped according to the number of ecotypes of the originating subset, i.e. to their degree of ecotype mixing, and the median for each quantile and number of ecotypes was given. From this point onward ‘quantiles’ refers to the median of these quantiles from the sampling procedure. The 0.50 quantile, i.e. the median, means that yield is higher than the quantile value with a probability of 0.5 (thus in 50% of cases). The 0.05 quantile means that yield is higher than the quantile value with a probability of 0.95. For P. nigra with eight ecotypes this resulted in 255 subsets belonging to eight ecotype richness groups and for F. sylvatica and the artificial data 127 subsets occurred which belonged to seven ecotype richness groups. The number of data points within the groups differed according to the number of possible combinations. The effect of this was attenuated by the method of calculation of the quantiles. Lottery sampling For lottery sampling one random individual was selected out of five randomly chosen individuals instead of the best performing. This simulation procedure was considered as an ecological and methodological ‘null-model’, although we did not test the model for significance but only for the effect that the process maximum 139
Manuscript 4: Ecotype mixing as climate change adaptation selection had (Gotelli 2001). Ecologically, lottery sampling means that survival of self-thinning and yield are uncorrelated. This can occur due to a lack of correlation between performance and self-thinning or due to a lack of correlation between performance and yield. Methodologically, simulations with lottery sampling give the same results as simply mixing the distributions (individuals representing the distributions) of the subsets for each group and calculating quantile values. In case subsets consist of only one ecotype, i.e. no ecotype mixing, the resulting quantiles are the same as the quantiles of a mixed distribution consisting of the single-ecotype distributions. This relates directly to the portfolio effect. Minimum selection In order to develop an ecologically ‘worst-case’ model, or a methodologically inversed model, we selected not for the best performing individual (maximum selection), but for the worst, i.e. we selected for the minimum (minimum selection). This represents a negative correlation between performance and self-thin- ning or a negative correlation between performance and yield (which is considered unlikely). A negative correlation between performance and self-thinning might occur if other factors than competition (mediated by performance) are the main causes of mortality such as herbivory or other disturbances (Franklin et al. 1987, White and Jentsch 2001, Dietze and Moorcroft 2011). Results Maximum selection resulted in increasing medians for yield with an increasing number of ecotypes, particularly for datasets with high variability among ecotypes and presence of over-performers. Inter-quantile distances increased along the gradient of ecotype richness, resulting in asymmetrically higher chances for high yield in more diverse stands (Fig. 1). The artificial dataset A, with normally distributed median performances and low variance between the ecotypes, resulted in almost no changes of median yield with increasing number of mixed ecotypes. A higher variance between ecotypes, like in dataset B, led to a stronger increase of the median yield with increasing ecotype richness in one stand. Inter-quantile distances were asymmetric to the benefit of higher yield the more ecotypes were combined. Adding one over-performer to the low variance dataset (dataset C) caused a steep increase of the upper quantiles (0.5, 0.75, 0.95 and maximum) of expected growth response from subsets with three ecotypes to subsets with four 140
Manuscript 4: Ecotype mixing as climate change adaptation ecotypes, with a slight decrease of the median with five or more ecotypes included in the subset. Adding one under-performer did not change these patterns considerably (dataset D). Dataset E, with two over-performers, exhibited a steep increase of expected growth from the single-ecotype subset to subsets with two ecotypes and a stable performance from that point on. Again, adding one under-performer did not change the pattern (dataset F). Maximum selection of the experimental data for F. sylvatica showed an increase of yield from subsets with one ecotype to subsets with two ecotypes for the quantiles 0.05 to 0.75. The expected maximum growth (0.95 and maximum) increased with increasing number of ecotypes. Subsets with more than four ecotypes did not yield additional positive effects on expected growth maxima. For P. nigra, with an increasing number of ecotypes, only the upper quantiles (0.95, maximum) increased. However, the increase in expected growth maxima (maximum, 0.95) culminated when 5-6 ecotypes were included. More ecotypes did not have additional positive effects. All other quantiles (minimum, 0.05, 0.25, 0.5, 0.75) remained remarkably stable, independent of the number of ecotypes included. Experimental data for P. nigra and F. sylvatica thus showed patterns similar to the artificial datasets C ‘one over-performer’ and E ‘two overperformers’. This indicated that the variability among ecotypes was large enough to show an effect for ecotype mixing and that it was not as small as in the artificial dataset A where the effect was not detectable (within versus between ecotype variability). However, P. nigra was a rather low variability dataset with only one or two weak over-performers (ecotypes c and e; boxplots in Fig. 2). Under maximum selection, the increase in median profit was therefore low, but still detectable. The dataset of F. sylvatica contained over-performers (ecotype a and e) and showed a clear increase of yields with increasing ecotype mixing. For the lottery sampling, the medians for yield remained unchanged with an increasing number of ecotypes. Inter-quantile distances increased, implying stronger effects with increasing ecotype richness. The influence of under- and over-performers was directly reflected in the quantile lines. Over-performers led to asymmetric patterns in the direction of high yield, under-performers in the direction of low yield (because the mixed distributions were skewed). The effect of statistical averaging causing the portfolio effect appeared in Fig. 1 and 2 by the comparison of the median of the lottery sampling for growing ecotypes separately, 141
Manuscript 4: Ecotype mixing as climate change adaptation 142 Fig. 1: Artificial data sets. A) Low variance, B) variance increased by factor ten, C) low variance data set with one over-performer, D) with one over-performer and one underperformer, E) with two over-performers and F) with two over-performers and one under performer on the right side, and on the left side the corresponding results of the simulations. Lines represent the medians of the quantiles (0, 0.05, 0.25, 0.5, 0.75, 0.95, 1) of 5000 replicates grouped according to number of ecotypes included.
Manuscript 4: Ecotype mixing as climate change adaptation i.e. single-ecotype stands, with the medians of the data boxplots. A change in the median performance of ecotype g between the datasets A and C only had a weak influence on the median of the simulated performance (Fig. 1). The results for the experimental data for F. sylvatica and P. sylvestris differed from the artificial data regarding the lower quantiles because they were truncated at zero. Contrary to the artificial data, lower quantiles were therefore close to zero and did not change with ecotype mixing. For minimum selection, inverse patterns regarding under- and over-performers compared to the maximum selection were observed, i.e. medians for yield decrease with increasing ecotype richness. In the artificial datasets B and D (containing under-performers), minimum selection resulted in a strong reduction of median yields (Fig. 1). Over-performers had almost no effect, only in dataset E with two strong over-performers. For the results of the experimental data the same was true as under lottery sampling, i.e. lower quantiles were close to zero and did not change. In the introduction it was argued that optimal yield could be reached when the ‘best’ ecotype is chosen (and grown without mixing). Therefore, the simulation results for single-ecotype stands with the highest median yield are also shown for each simulation (Fig. 1; ‘Best’). It appears that for maximum selection the optimal 143 Fig. 2: Data and results of the three selection methods for P. nigra and F. sylvatica. Lines represent the medians of the quantiles (0, 0.05, 0.25, 0.5, 0.75, 0.95, 1) of 5000 replicates grouped according to number of ecotypes included. Additionally the same quantiles for the best performing ecotype are given.
Manuscript 4: Ecotype mixing as climate change adaptation choice actually led to higher yield than mixed stands. Also for the experimental data choosing the best ecotype resulted in higher yield (Fig. 2; ‘Best’). This was also true for lottery sampling and minimum selection, where the difference was even larger. Discussion The present simulation study aids to understand the effects of self-thinning and statistical averaging (i.e. the portfolio effect) on expected yields applying ecotype mixing. We hypothesized that the interplay of these two processes linked by performance might exempt the portfolio effect from its drawback of lower chance for high profits (Hopper 1999, Boyce et al. 2002, Childs et al. 2010) while the advantage of stabilization remains (Markowitz 1952). If this is the case, ecotype mixing constitutes an opportunity for the adaptation of forest stands to present climate change, above all to prevent possibly high yield losses. Although we only explore intra-species diversity in terms of ecotypes, these results also apply to the species level (i.e. mixing species instead of ecotypes). In the present study, two different plantation strategies are addressed as opposed to betting on the apparently best ecotype. The first strategy of growing ecotypes in separate stands is directly related to the classical portfolio effect with its advantage of stabilization of products compared to betting on the assumed best ecotype under high uncertainty. In this strategy, single-ecotype stands can be considered as assets of a portfolio of stands. For the portfolio effect the expected yield of each single asset is compared to the expected yield of the bundle of assets. The expected yield of the portfolio has a lower variability than the expected yield when only one asset with unknown performance was chosen. However, the opportunity provided by the process of self-thinning does not apply for this planting strategy as self-thinning would only take away individuals of the same ecotype. The second planting strategy is ecotype mixing. Only when ecotypes are actually mixed in the same stand do they compete, and the beneficial effect of selfthinning can occur (individuals as assets, stand as portfolio). However, mixing of ecotypes results in mixed distributions. Assuming lottery sampling, expected yield in such stands is the same as the expected yield for the portfolio of single-ecotype stands. However, the variance of a mixed distribution is larger than the variances of the single distributions in case their means differ (Boes 1966).This is the reason 144
Manuscript 4: Ecotype mixing as climate change adaptation for the divergence of the quantile lines representing the probabilities to obtain a yield equal or lower than that value, with increasing ecotype richness (Fig. 1 and 2). The uncertainty of the median yield is therefore higher for the strategy of ecotype mixing than for a portfolio of single-ecotype stands. Thus, in case the reaction of ecotypes is not known, and under the assumption of lottery sampling, the median yield is not stabilized by ecotype mixing as compared to singleecotype stands (finding 1). However, under the assumption that both, selection and yield are determined by performance, ecotype mixing leads to higher median yield than a portfolio of single-ecotype stands (maximum selection). This is due to self-thinning removing low performance individuals from the stands. By this, especially low quantiles are increased considerably as compared to lottery sampling and minimum selection. The more over-performers are among the ecotypes, the earlier the median increases (probability larger than 0.5 that over-performer is in the mixture). If maximum selection is an appropriate model for the interplay of performance, selfthinning and yield, ecotype mixing increases the chance for high yield but does not increase the risk of low yield despite an increase of within-stand variability. However, yield is not as high as it could be under the optimal choice of the best performing ecotype. The portfolio effect is thereby not entirely but largely exempted from its drawback of lower chances for high yields (finding 2). Patterns for the experimental data are similar to the artificial datasets with one and two over-performers and show an increase of expected yields under ecotype mixing. This points to the opportunity that the intra-species variability for F. sylvatica and P. nigra have a structure, which results in a positive effect of ecotype mixing (for low variability datasets ecotype mixing has no effect). The results for the experimental data differ from the artificial data because they have a lower boundary of growth (zero). Under-performers are therefore limited and the lower quantiles change less with ecotype mixing. The other patterns described above are found for the experimental data, especially for F. sylvatica while the effects are rather small for P. nigra. This is caused by the lack of strong overperformers in the P. nigra dataset. Contrary to the data used in the present study, former studies have found high intra-specific differences on a genetic level not only for F. sylvatica (Konnert 1995, Vornam et al. 2004) but also for P. nigra (Aguinagalde et al. 1997, Varelides et al. 2001, Jagielska et al. 2007, Lucic et al. 145