Developing a physiologically mechanistic tree migration model and simulating Holocene spread of forest trees

Heike Lischke¹, Brigitta Ammann², David W. Roberts³, Niklaus E. Zimmermann^1,3

¹ Swiss Federal Institute of Forest Snow and Landscape Research, CH-8903 Birmensdorf, Switzerland; ² Institute of Plant Ecology, University of Bern, CH-3013 Bern, Switzerland; ³ Dept. of Forest Resources and Ecology Center, Utah State University, Logan, UT 84322-5215, USA

Project Summary

We propose to develop a new, spatially explicit, physiologically mechanistic, and efficient forest succession model to study the potential of trees to invade, migrate and get extinct under current and possibly changed future climate. The potential of plants to migrate to habitats which are more favourable under changed conditions plays a key role (Kirschbaum & Fischlin 1996) in assessing the impacts of climate change on terrestrial ecosystems.
To study such migration potential a combination of a) dynamic ecosystem simulation studies, b) prehistoric data on species composition and distribution, and c) independent climate archives is the most promising approach. Such studies require high-resolution, high-quality data and a biologically trustworthy, efficient, and spatially explicit forest succession model that includes a seed dispersal module.
To achieve this, we propose to combine the power and strengths of two different, previously developed models, each suitable today for parts of the required task only. Major research will be conducted in the field of implementing and calibrating a seed dispersal module that enables to simulate short- and long-term migration rates of trees under realistic forest dynamics. Once developed it will be tested with palynological data available for Switzerland.
Keywords: Tree species migration, Forest modelling, Paleoecology, Climate change

1. Introduction

This project is an ongoing effort to conceptually develop, code, and parameterize a physiologically mechanistic and spatially explicit forest succession model. We propose to make use of two different models (LAASIM and DisCForM), which are optimized in separate aspects each(Fig.1).

Fig1: Illustration of the proposed study. The main stages in this project are model development and model evaluation

The propotype of the LAASIM model was developed in 1993 (Roberts et al., 1993). The initial version was based on hypothetical tree species and was applied to hypothetical landscapes. Currently, the model is applied to real landscapes and is parameterized for real tree species of the Central Rocky Mountains. Also, the concept of the model was slightly adjusted. The strength of the LAASIM model is its clear conceptual basis and the physiological mechanisms it is based on, and the parsimonious programming.

The DiscForM model was developed recently (Lischke et al., 1998c). The model is based on an aggregation approach to simplify and accelerate the calculations of complex population dynamical processes without sacrificing accuracy. It thus can be viewed as a highly optimized model for the task of spatially explicit calculations of forest succession. Furthermore, the model has been tested successfully against palynological records (Lischke et al., 1998a).

The projects starts June, 1^st 1999, and is funded by the Swiss National Science Foundation. The final report is due May, 31^st, 2001.

2. State of the Art in modelling tree dynamics and migration

The recent (Santer et al. 1996) and projected future change in the global climate (Houghton et al. 1996) may have significant impacts on the terrestrial vegetation (Watson et al. 1996). Of all terrestrial plants trees are one of the best investigated groups and they play the most important role in the global carbon cycle (Kirschbaum & Fischlin 1996), accounting for approx. 80% of terrestrial carbon (Dixon et al. 1994). Thus, we will focus on the dynamics and spatial patterns of trees. A changing climate alters individual tree’s living conditions, recruitment, survival, and competitive abilities, and by this the species composition and dynamics at a given location.

On the other hand tree species can adapt to changes by migrating through seed dispersal to new habitats where the site conditions are more favourable to them. This reflects an adaptation towards the balance of cen-trifugal (seed dispersal) and centripetal (environmental control) forces (Sauer 1988) that may or may not reach an equilibrium, depending on the stability of the environmental conditions and of the competitive structure of the forest communities.

The potential and speed of such tree species’ migration is crucial for the assessment of how ecosystems re-spond to changes in climate as fast as currently predicted (Solomon & Kirilenko 1997). It determines e.g. whether a forest with its present species will:

change its component tree species through invasion of better competitors or more stress tolerant trees (de-pending on the nature of the change) by competitively replacing the less adapted species.
reveal mainly higher or lower stand growth (depending on the nature of the change) with an associated minor change in the dominance of present species because other, better adapted species are unable to migrate (fast enough) to this site predicted (Solomon & Kirilenko 1997).
will break down and be replaced by shrubs or grasslands because the new site conditions are no longer suitable for

the present trees and/or more stress tolerant species are unable to migrate (fast enough) to this site, as is discussed e.g. for the continental climate of Valais (Bugmann 1997).

This combined effect of forest dynamics, climate variability, and tree migration on the spatio-temporal pattern of species abundance has been shown in various paleoecological studies.

Paleoecologists have long recognised that plant species composition has changed significantly over time. Past variation in climate is discussed as the main reason for the resulting changes of species distribution in geo-graphical space (Davis et al. 1986; Kloetzli et al. 1996; Lischke et al. 1998a; Prentice 1986; Ritchie 1986; Webb 1981), as well as for their altitudinal distribution, e.g. near timberline (Gear & Huntley 1991; Grabherr et al. 1994; Kullman & Engelmark 1991; Lotter et al. 1998; Wick & Tinner 1997). It is now well-established from many paleo-ecological studies that modern species assemblages do not have long histories (Birks 1993; Davis 1983), that dispersal of seeds is a major factor in determining how fast trees can migrate (Roberts 1989), and that due to the individualistic migration rates (Bennet 1983) there are lags in the readjustments of vegetation composition (Davis 1989).

Since palynological data can be resolved at the generic or even species level for most of the temperate and boreal tree species (Bennet & Lamb 1988), a sound paleoecological data basis exists to assess the migration potential of trees.

Several simulation studies exist where for single, isolated locations, simulations from forest models have been compared to vegetation patterns derived from prehistoric pollen records. Solomon and co-workers (Solomon & Bartlein 1992; Solomon et al. 1981) tested the simulations with the model FORET against North American pollen records; climate and immigration scenarios were inferred from one of these records. Lotter and Kienast (1992), Fischlin et al. (1995b), and Lischke et. al. (1998b) used an annually varved pollen record (Lotter 1989) from Central Switzerland to verify the successional pattern produced by the models FORECE (Kienast 1987) and ForClim (Bugmann 1996), given a sequential appearance of the tree taxa in the catchment area and various climate scenarios. The dates of appearance were derived from the pollen record itself.

Even though these comparisons reveal reasonable coincidence between predicted and observed presence and abundance of trees under various climatic conditions, these studies do not test the predictions of invasion and migration of trees in space and time, mainly because a) the indications of past climatic change are derived indirectly from the arrival of new trees (and thus circumvent the problem of migrational lags), and b) the studies are performed for isolated locations only. Such tests have to be performed on a series of locations simultane-ously, a task that is now feasible thanks to the establishment of large palynological data bases such as the pollen database for the European Alps (van der Knaap & Ammann 1997).

2.1. Spatial forest simulation studies

There exist some studies of spatial vegetation simulations including dispersal. Dyer (1995) studied migration for a wind- and an animal- dispersed species on a cellular automaton with landscape barriers, but neglected forest dynamics. Some forest dynamic models were adapted in order to predict the forest dynamics spatially explicit including seed dispersal subroutines. Examples include the SORTIE-model (1996; Pacala et al. 1993), a very detailed approach that runs on small areas only, the LANDSIM- (Roberts 1996a; Will-Wolf & Roberts 1993), and the LANDIS-model (Mladenoff et al. 1993). Both latter models are fast and simple forest succession models, based on the vital attribute concept (Noble & Slatyer 1980), and suitable for running on regional spatial scales. However, the two models are not developed from physiologically mechanistic principles, and simulation studies along entire migrational paths including forest dynamics have not yet been carried out to our knowledge. Although fast and simple, these two models have only limited capability to simulated tree species response to transient climate, since they parameterise the realised rather than the fundamental niche of the model trees. Under rapid warming scenarios such models tend to simulate large-scale dye-backs of forests, instead of al-tered growth patterns and adjusted migration rates.

Thus, in order to predict the tree species’ potential to respond to climate change by migration, seed dispersal, forest dynamics and climate influence have to be investigated simultaneously. For this aim the combined use of dynamic models which are based on physiological and ecological processes (in particular seed dispersal), of climate scenarios, and of tree migration data is the most promising approach.

2.2. Seed dispersal

Seed dispersal is considered as the primary driver for plant migration (Sauer 1988), besides the local forest dynamics along migrational paths. Seed dispersal depends on the reproductive traits of a species, such as minimum reproductive age, the amount of seeds produced, and on the distances and probabilities of seed transport by the vector. This distance-probability relationship is generally parameterised in models through seed-rain curves. There exists a wealth of data characterising seed dispersal of north-American species (Brown et al. 1988; Chambers & McMahon 1994; Farmer 1997; Green 1980; Greene & Johnson 1989; 1995; Hutchins & Lanner 1982; Johnson & Adkisson 1985; Matlack 1987; Stapanian & Smith 1986; van der Wall & Balda 1977) particularly for wind-dispersed species. For (Central-) European species such data are equally available, although less abun-dant (e.g. Jensen 1985; Matlack 1987; Stoecklin & Baeumler 1996).

Species specific seed rain curves determine the local pattern of seed availability and regeneration and influence the patterns of forest composition emerging from forest dynamics (Pacala & Deutschman 1995). Thus the introduction of seed dispersal is a serious improvement over the assumption of non-spatial models that at any given location in a landscape all tree species’ seeds are equally available (Finegan 1984). On large scales seed availability strongly influences the expansion of species along environmental or species composition gradients, i.e. migration.

The way, seed dispersal curves are take into account in a model appears to be crucial. It is not clear which features and part of seed rain curves are responsible for the resulting migration speed at different spatial scales: the initial part, characterising the frequent short distance dispersal, or the tail of the distribution, indicating rare long distance seed transports (Cain et al. 1998; Williamson 1996). Average seed transport distances alone are not sufficient to explain migration rates. Even if available for all species of interest, there remains a discrepancy between measured dispersal distances and migration rates as observed from palynological studies (Cain et al. 1998; Greene & Johnson 1995; Skellam 1951; Webb 1986). To explain such discrepancies, the distinction be-tween short- and long-distance dispersal is proposed (Hengeveld 1989; Shigesada & Kawasaki 1997). However, little is known about the importance of the tail of seed rain-curves (Cain et al. 1998; Malanson & Armstrong 1996; Portnoy & Willson 1993). The highly stochastic, but extremely rare long-distance dispersal events are beyond the scope of experimental studies (Cain et al. 1998), while short-distance dispersal of seed-rain is easier to assess and to describe with deterministic seed-rain curves. However, long distance dispersal might be as-sessed (Lischke et al. 1998a) by recorded maximum transport distances (Hughes et al. 1994) or properties of the seed-vector, e.g. the radius of action of seed transporting birds (Johnson & Webb III 1989; Stimm & Böswald 1994).

2.3. Forest Models

Studying tree species migration mechanistically requires a forest model which fulfils several basic requirements.

It has to be based on sound biological and ecological knowledge about processes and relationships in the system forest, rather than on simple empirical (statistical) relationships,. The fundamen-tal instead of the realised niche of trees has thus to be parameterised. By this, it is more likely to exhibit a reasonable behaviour also under conditions never experienced during the time periods the data used to derive empirical relationships stem from.
It must be able to capture the effects of variability caused by stochastic birth and death processes of single individuals and temporally fluctuating environmental input.
Its parameters have to be valid for Central European tree species.
It requires a seed production-, a seed dispersal module, and the capability to simulate the regen-eration of trees proportionally to the locally available seeds.
It must be implemented spatially explicit, i.e. on a grid, and be able to communicate with a GIS.
It must be fast enough for the numerous simulations required for spatial simulations.

So far, ecologists have developed a variety of forest models ranging from relatively simple Markov chain models (Horn 1975; Moore 1990), ”vital attributes” conceptual models (Catellino et al. 1979; Noble & Slatyer 1977; 1980; Roberts 1996a, 1996b), and abstract mathematical models, such as the height-structured population models (Karev 1994; Kohyama 1993; Kraev 1998; Saldana 1998) to more detailed simulation models. A wide-spread group of forest models are gap models (Botkin et al. 1972a; 1972b; Bugmann 1996; Kienast 1987; Pacala et al. 1993, 1996; Pastor & Post 1986; Prentice et al. 1993; Shugart 1984; Smith & Urban 1988; Urban & Shugart 1989), stochastic models, which follow the establishment, growth and death of individual trees on small patches. Gap models have been extraordinarily successful and were applied under current conditions and in climate change impact studies (e.g. Bugmann & Fischlin 1996; 1996; Fischlin et al. 1995a; Kraeuchi & Kienast 1993; Prentice et al. 1993; Solomon et al. 1981; Urban & Shugart 1989). However in order to improve our under-standing of large-scale dynamics under current and possible future environmental conditions, we believe that it is more appropriate to conceptualize a new forest simulation model from physiological first principles.

We agree with Bonan & Sirois (1992) and Loehle & LeBlanc (1996) that while many gap-based climate change models are relatively successful in portraying species distributions under current climatic conditions, they are often successful for the wrong reasons, assuming that the observed biogeographical distribution of species reflects the fundamental relation of the species to environmental conditions, whereas the distribution is also determined by biotic processes such as competition or migration. The parameters of gap models are not based on physiological experiments, rather on empirical observations (longevity, maximum size, geographical distribution, etc. of trees). While some of these parameters reflect the fundamental niche, some other parame-ters are an expression of the realised niche instead (e.g. the upper limit of daydegrees). The resulting model can be described as quasi-mechanistic.

Alternatively, detailed physiological simulation models such as Forest-BGC (Running 1994; Running & Coughlan 1988; Running & Gower 1991) generally lack the compositional and structural information about vegetation that is necessary to realistically depict forest dynamics (Deutschman 1998; Lischke et al. 1998c). In addition, they operate on time scales that are simply too short to allow us to model large landscapes for long periods of time. Detailed physiological stand growth models on the other hand, which focus on individual trees are very slow and confined to single locations or small areas (e.g. Bossel 1987; 1991; Bossel et al. 1991; Deutschman et al. 1997; Pacala & Deutschman 1995; Pretzsch 1992, 1995). However, such models serve as an integration of our understanding of basic ecological processes at a range of temporal/spatial scales and levels of organization.

As an alternative, currently gap models are developed which are based on fundamental ecological and physio-logical knowledge. An example is the gap model 4C (Bugmann et al. 1997), which includes a mechanistic rep-resentation of photosynthesis, carbon allocation patterns and disturbance regimes. However, this model is still based on the time-consuming stochastic gap-model approach and it is not spatially explicit.

2.4. Aggregation of forest models

The opposite to such an top-down approach of expanding a simple model by introducing more physiology-based components is the bottom-up approach of aggregating detailed models in a controlled way to simpler ones.
Ecological simulation models which are based on detailed physiological and ecological processes are generally very slow and error-sensitive. Furthermore, they are often not mathematically tractable. Classical mathematical analyses, such a equilibrium-, stability-, sensitivity- or bifurcation- analyses, which can give essential insights into the system, are thus impossible.

Several approaches exist to aggregate components of detailed models (Iwasa et al. 1987; 1989), e.g. the sepa-ration of time scales (Auger & Roussarie 1994). In forest modelling the approximated moment equation (Bolker & Pacala 1997) has been applied to aggregate the position dependent model SORTIE (Pacala et al. 1993, 1996) to a structured population model. Kohyama and co-workers (1989; 1992; 1993; 1989) used a diffusion-equation approach to describe stand dynamics, given mean and variance of individual tree growth rates de-pending on the cumulative basal area of the stand. Fulton (1991) aggregated the individual based gap model FORSKA (Leemans & Prentice 1989) to the stochastic structured population model FLAM by introducing dis-crete height classes, however keeping the stochastic replicates of gap dynamics. Kraev (1998) formulated the gap model ForClim as a partial differential equation, depending on time and height. Recently Kirkilonis (1998) proposed an approach of individual trees competing via interaction-circles, which he aggregated also based on an approximated moment-equation.

None of these approaches fulfills all requirements for a tree species migration model listed above. We believe that a combination of the mentioned approaches, namely the aggregation of a physiologically mechanistic, individual-based model is necessary, to obtain a manageable, trustworthy migration model.

3. Research contribution of principal investigators

3.1. Pollen and independent climate reconstructions

Pollen: For the validation of ecological models, data of adequate temporal and spatial resolution are essential. Because of the low rate of tree migration, pollen records are the best choice. Quality of data depends on their spatial and temporal resolution: for Switzerland the Alpine Palynological Data Base (ALPADABA, supported by NF grant nr. 31-37620.93) contains original raw data of more than 100 sites, and many of them are well dated (van der Knaap & Ammann 1997). Special efforts were made in this project to improve the time control by re-dating sites by both AMS-techniques (58 dates) and decay-counting (31 dates). New studies (e.g. Wick 1996; Wick & Tinner 1997) and current projects are providing additional results (NF-3100-923 for the Southern Alps; NF-21-49606 for the Engadin; NF 1214-047092.96 for the Swiss Plateau; and for the Canton Zug). At some well dated sites presence or absence of tree genera can be assessed by influx values (pollen per surface area and time).

The model validation to be developed in this project will be applied to the continental scale of the European Pollen Data base (EPD) – this is ensured because B. Ammann was in the executive committee of the EPD since its beginning (1989-1997) and is now replaced by Dr. André Lotter, also of Bern. Full compatibility between ALPADABA and EPD was always guaranteed.

Plant Macrofossils: Seeds, fruits, needles, etc. are important in assessing the local presence or absence of tree taxa independent of pollen, which can be easily transported from afar. Only a few sites in the entire Alps have been analysed in the necessary detail for their plant macrofossils. Some of the best ones are in or very near Switzerland (e.g. Tinner et al. 1996; Wick 1994, 1996). For conifers a good first approximation to macroremains are the stomata (Ammann & Wick 1993); they were recognized and counted by many early workers (e.g. Welten 1982), and their records are now stored in ALPADABA.

Climatic reconstruction independent of tree pollen: Climatic reconstruction in itself would be too big a task to be included in this project. However, for selected time windows the attempt is feasible on the basis of multi-proxy approaches: either with water plants (Ammann 1989; Haas et al. 1998) or through our experience in co-operation with colleagues working on insects (Ammann et al. 1983, 1985; Elias & Wilkinson 1983; 1985), aquatic invertebrates (Lotter, in prep. Lotter et al. 1997; 1998), oxygen-isotope ratios (Ammann et al. 1998; Eicher & Siegenthaler 1983), or combinations of these.

3.2. Paleoecological simulation study

To better understand whether the migration of Central European trees in the past was controlled primarily by climate or by dispersal and regeneration abilities, Lischke et al. (1998a; 1998b) examined why tree taxa were temporarily absent in a Holocene pollen record in the Central Swiss Plateau. Simulations with the forest gap model ForClim (Bugmann 1994, 1996) were compared with a high resolution pollen record from the Swiss Plateau. Several combinations of transient climate and of tree species immigration scenarios were applied successively, to systematically test the effects of migration(al lags), direct climate influence, and competition. The temperature and precipitation inputs were reconstructed from largely independent data based on cladoceran, aquatic plant remains and oxygen isotopes in sediments and on lake level measurements to avoid circular effects between input data, model, and comparison data. The results suggest that during late Würm delayed immigration played an overwhelming role, whereas during and after Younger Dryas temperature and precipitation influenced the pollen record both directly and indirectly by changed competition. For two taxa, Abies and Fagus, we concluded that their immigration was controlled by their regeneration and dispersal abilities or by geographical barriers because they appeared later in the pollen record than climatic conditions would allow. Their earliest arrival times were inferred from regional and continental scale pollen maps. These arrival times corresponded well with arrival times which had been assessed based on the distances to glacial refuges, the length and the climate conditions of potential migrational paths, and on maximum migration speeds of these species. The latter had been estimated from the minimum regeneration time and maximum dispersal distances, which are recorded for the seeds or potential vectors.

3.3. Aggregation of a forest gap model

The paleoecological simulation study (Lischke et al. 1998a) revealed the importance of assessing the tree migration for climate-change studies. It also demonstrated that gap models are not suitable for migration simulations: A single run of ForClim (Bugmann 1996) for one observation site and the period of 7000 years took more than one hour on a SUN Ultra-Enterprise. For this reason Lischke et. al. (1998c) developed an approach to aggregate the semi-empirical gap model ForClim to the distribution-based approach of DisCForM (Lischke et al. 1998c). DisCForM is a height-structured population model. Instead of describing the dynamics of several properties of each individual tree, it describes the dynamics of the population densities of trees in each tree height class. Its fundamental approach is to describe the variability of tree population densities from patch to patch by theoretical frequency distributions. In gap models the same result is achieved by numerous, time consuming replicates of stochastic simulations. The aggregation approach results in modelling distributions of light intensities and consequently of light-dependent process rates such as growth, birth and mortality. This produces a purely deterministic description of the dynamics which still reflects the variability in a forest.

Besides this structural difference, DisCForM still resembles ForClim. It uses the same process functions for establishment, growth and death, including their dependencies of environmental factors. Consequently, the simulation results are very similar to those of ForClim, however, the model needs a drastically shorter (5%) simulation time, which makes it suitable for large-scale applications. Since ForClim has been validated in many locations and DisCForM behaves similarly, the aggregation approach can be considered as trustworthy. Additionally, validations of DisCForM against Swiss National Forest Inventory Data originating from various climatic conditions proved this assumption successfully (Lischke 1998; Lischke et al. 1998c; Löffler & Lischke 1999). Not only is this approach computationally much more efficient than the gap-model approach, but it is also mathematically transparent: It consists of a system of ordinary differential equations, one for the population density of each species in each height class.

Currently, a module to incorporate also temporal variability of climate input (Löffler & Lischke 1999) and to simulate seed production and sapling development are under way.

This model fulfils requirements 2, 3, and 6 and in parts 1 and 4 (see 2.3.).

3.4. A physiologically mechanistic forest simulation model

In a prior project we have developed, coded, and debugged a spatially explicit, physiologically mechanistic simulation model (Roberts et al. 1993; Zimmermann & Roberts in prep.).

The model simulates establishment, growth and death of individual trees at specific locations. It first determines the maximum possible leaf area at given site conditions and proportions the relative growth ofspecies and individuals thereafter, based on available resources, and the relative canopy position of individuals.

The model is based on fundamental relationships between (1) soil moisture and stand leaf area (Gholz 1982; Grier & Running 1977), sapwood cross-sectional area and tree leaf area (Margolis et al. 1995; Waring et al. 1977) to simulate the dynamics of leafs. Leaf area is the primary driver to simulate photosynthetic carbon gain. The resulting growth is influenced and controlled by the amount of shading, of thermal energy, by limitations in nutrient and water availability, and by stress through low temperature, thus reflecting sensitivity to physiologically indispensable resources. The response of trees to these parameters is based on the theoretical framework of the trade-off theory (Smith & Huston 1989). This theory assumes that trees (and plants in general) have adapted evolutionarily by either optimising stress-tolerance or growth under optimal resource availability. This reflects a basic trade-off in the use and allocation of photosynthetically allocated carbon. The trade-off model demonstrates effects of limitations in resource availability on the rate of growth and, thus, makes predictions about the fundamental niche of plants. One of the key problems of this approach is how to partition the total amount of leafs that can be allocated in a stand among trees. We solve this by tree leaf area-site water balance functions, reflecting the LAI a tree can develop under various moisture regimes. The species specific maximum is defined by the shade-tolerance of the tree while the minimum of this asymptotic curve is given by its drought tolerance. The theories and concepts the model is based on are reviewed more detailed here.

We have tested the model on hypothetical species for hypothetical landscapes. The model has been shown to exhibit: (1) realistic successional dynamics, (2) realistic leaf area dynamics, and (3) appropriate sensitivity to climate and soil water holding capacity. Simulated changes in soil parent material or topographic position influence soil moisture holding capacity, and the model responds with appropriate changes in species composition. Simulated changes in climate also produce realistic vegetation transient responses with vegetation inertia on the order of decades to centuries and re-adjustments in equilibrium distribution that appear reasonable (Roberts et al. 1993).

We have continued the model development in the last year. First, we have modified several of the physiological mechanisms in the model to reflect an increased understanding of sapwood area/leaf area ratios, leaf area/site water balance ratios, and individual tree characteristics. Second, the proposed model is currently calibrated for real species instead of the hypothetical species. Third, the proposed model is now spatially explicit with GIS-oriented in- and output, to simulate vegetation dynamics on a real landscape. These changes accomplish two major improvements: (1) the modelled landscapes have realistic patterns of spatial auto-correlation and heterogeneity, and (2) most important, the model is subject to validation and suitable for testing a large number of specific hypotheses at all levels from individual tree to stand to landscape.

The model is driven by environmental variables that are derived from GIS-based data. In order to run the model for a specific area, we have developed a framework of tools to generate digital maps of bioclimatic variables within reasonable time.

Consequently, the model fulfils requirements 1, 2, 5, and partially 6 (see 2.3.).

4. Research Plan

A major part of the project will be devoted to develop and test a spatially explicit forest-simulation tool, including a seed-dispersal module. The required input data for present day simulations will be derived from biophysical maps. The model will be parmeterised and first be evaluated at single locations. Once a seed-dispersal module is developed the simulation of tree migration will become possible. The model will be applied to limited time windows in the Late Glacial and Holocene and compared with pollen data and macrofossils. The required independent climate input scenarios will be reconstructed from various data sources.

4.1. Simulation tool

Goal:
Develop a simulation tool to calculate spatially explicit tree migration based on mechanistically modelling species competition in space and time by combining the approaches of DisCForM and LAASIM.

Methods:
The fundamentals of the model
We propose to develop a new simulation tool based on two currently available models by incorporating their respective strengths, and by parameterising the simulator for Central European tree species. The proposed model is based on the physiologically mechanistic process functions of LAASIM (Roberts et al. 1993; Zimmermann & Roberts in prep.), and adds the economy of performance of DisCForM (Lischke et al. 1998c). Both models are currently designed to optimise performance speed, in order to be able to calculate large numbers of sites within reasonable time. The main simplification over stochastic Gap models is the fact that the dynamics of trees is simulated deterministically.

In a first phase we will assume that the seeds available in each simulated grid cell (landscape element) are homogeneously distributed. This allows to apply the original DisCForM aggregation approach. In later versions, we will aggregate the individual tree dynamics of stands where seed dispersal occurs also inside the grid cell, based on a combination of the DisCForM-approach with moment equation approaches (Bolker & Pacala 1997; Kirkilonis 1998).

DisCForM even goes a step further in optimising computational resources. Instead of modelling individual trees it lumps single stems to diameter classes. Together with the deterministic, distribution based representation of the process functions this results in a reduction of 95% of computing time compared to the original gap model (Lischke et al. 1998c).

A major improvement of LAASIM over gap models is the effort to parameterise the fundamental (physiological) rather than the realised (ecological) niche. This is important in so far, as the widely observable realised niche is mostly a function of equilibrium conditions (competitional and climatic). However, to be able to simulate forest dynamics under transient climate realistically, we need to be able to model the tree species behaviour more fundamentally, independent of former equilibria observations. Thus, we must attempt to parameterise the physiological instead of the ecological limits of trees, and to design the competitional model functions from physiological principles as rigorously as possible. Constraints apply due to the fact that the model is intended to run for long time series and on large landscapes.

Deterministic vs. stochastic modelling
While the majority of existing gap models simulates stand dynamics as stochastic processes in small canopy openings (gaps), we propose to develop the forest simulator based on deterministic functions, more similar to physiological ecosystem models (e.g. Running & Coughlan 1988). The rationale for simulating forest dynamics stochastically, and independently on small areas (~1/12 ha) is the general acceptance of the mosaic-cycle theory. By doing so, for each location forest succession for 50-200 gaps is calculated independently over several hundred years (each affected by stochastic disturbance over time), and averaged to get an overall view of what might be expected from a larger stand (several ha to km²).Lischke et al. (1998c) have demonstrated that the same pattern can be generated by a deterministic approximation of the individual response functions, which is based on assumptions about the spatial distribution of trees in the simulated region. When implementing a forest simulator for large spatial scales at a high spatial resolution, it does no longer make sense to simulate a large number of gaps on grid-cells barely larger than the assumed gap. Instead, we propose to benefit from the performance economy of deterministic,distribution basedresponse functions. This does by no means imply that chance events are unimportant nor that they can no longer be incorporated into such a model. We propose to simulate in a first version stochastic events dependent on biophysical and biological factors (monthly mean temperature, soil depth, insect density, proximity to rivers, slope angle, etc.). However, we aim at replacing also this stochasticity by a deterministic approximation.

Proposed Work:
We propose to develop a new spatially explicit forest-succession model based on the physiologically mechanistic functions of the LAASIM model. The new model will be computationally efficient and written in a mathematically closed form due to the aggregation approach applied and tested successfully in DisCForM. The model process functions are primarily based on the trade-off model, is specifically site water sensitive and simulates growth of trees as a function of photosynthetic carbon gain through leafs. The model is additionally sensitive to heat sum (energy) and nutrient availability, and simulates successional replacement of trees as a function of growth rates and shade tolerance.

4.2. Development of biophysical maps

Goal:
Develop high-resolution maps of biophysical input parameters for the study area. The maps are necessary to operate the model spatially explicit. Additionally, the maps are needed to calibrate some of the model parameters.

Methods:
A series of high-resolution climate maps is currently available for Switzerland (Zimmermann 1996; Zimmermann & Kienast 1995; in press). These climate maps are calculated in 25-50m spatial resolution which is sufficient for this modelling exercise. The available maps cover monthly mean values of average temperature (T), precipitation (P), solar radiation (R), potential evapotranspiration (ETp), and atmospheric water balance. The missing soil water availability maps are based partly on these available climate maps and on soil depth, coarse fragment content, and specific soil moisture holding capacity. The latter parameter can be derived (see Roberts et al. 1993) from the soil suitability map of Switzerland (EJPD-Bundesamt für Raumplanung et al. 1980), digitally available at FSL/WSL in Birmensdorf

Yearly available soil water (expressed as the site water balance) is one of the primary limiting factors to the development of stand leaf area, and thus to tree growth. We calculate the site water balance similarly to Grier & Running (1977) as the yearly sum of the monthly atmospheric water balance (P - ETp) over the soil bucket.

Proposed Work:
We propose to calculate additional maps of (1) site water balance and (2) minimum temperature for Switzerland at 25m of spatial resolution, according to the maps already generated for the same area. The site water balance will be calculated on the formulae presented above, while the minimum temperature is generated from SMI-climate stations data as described in Zimmermann and Kienast (1995; in press) and Zimmermann (1996).

4.3. Parametrization and evaluation of the model at single locations

Goal:
Parameterise the model components stemming from LAASIM for Central European tree species and run the model to climatic (dynamic) equilibrium for selected locations with known vegetation composition and forest structure to test the validity of the integrated logical consequences of the component theories and concepts the model is based on.

Methods:
Before the model is applied to large landscapes, it has to be calibrated and tested on a series of evaluation sites in order to ascertain that the model is capable of simulating the forest dynamics under various biophysical conditions. The calibration of forest succession models either is based on an analysis of literature (e.g. Bugmann 1994; Kienast 1987), based on an extensive field data set (e.g. Pacala et al. 1993, 1996), or on a combination of both (Zimmermann & Roberts in prep.). For the proposed model, a number of parameters can be derived from literature and from existing succession models on Central European forests (Bugmann 1994; Lischke et al. 1998c). Such parameters include the maximum age, maximum diameter at breast height (DBH), maximum basal area at breast height (BA), number of years to reach maturity, number of years to produce viable seeds. Other parameters have to be derived from analyses on extensive forest inventory data sets, such as the Swiss National Forest Inventory (NFI) (EAFV 1988; WSL 1998) database.

The Swiss NFI allows to derive a number of model parameters. It is sampled on a 1km-grid over the whole Swiss territory resulting in ~12'000 forested plots. Each plot was sampled on a 500m² and on a 200m² circle, where a series of stand structural and stand compositional data was recorded. It is planned to repeat the measurements approximately every ten years. So far, two sampling procedures have been performed (1983-1985 and 1993-1995). The recorded data can be split up into individual species’ characteristics. Overlaid with biophysical maps, a series of ranges and range limits along biophysical gradients can be derived from. Specifically, it allows to derive the following necessary parameters for each modelled tree species:

· lower range limit of site water balance
· lower range limit of heat sum
· lower range limit of minimum temperature By deriving the lower observed limits of resource gradients only, we will derive parameters that are close to the physiological limits of the respective trees, we thus assess elements of the fundamental niche (see the trade-off model in the LAASIM concept for further explanations, or Prentice et al. [1992] for a model example).

The most crucial task will be to derive the maximum LAI a tree can generate. The highest LAI under given soil moisture conditions are defined by the physiological shade tolerance of a tree species. Again, the trade-off model (Smith & Huston 1989) predicts that the maximum LAI that can be observed for a tree species is an expression of its stress tolerance, or a function of its physiological limits towards low resource availability (Bazzaz 1979; Boardman 1977). We will derive these parameters from LAI measurements at long-term monitoring plots that have been sampled by several projects at WSL during the last two years.

Once the calibration is completed, we will test the model under various biophysical conditions. This is important to guarantee that the model is able to simulate reasonable successional, structural, and compositional patterns under a range of different environmental conditionsbefore it is applied to a larger landscape, and under transient climate. Again, the NFI-database, in combination with data from long-term monitoring plots, is an invaluable source of evaluation data. We propose to select classes of evaluation plots in the biopysical factor space in a stratified selection procedure in order to ascertain that the following variables are appropriately sampled:

site water balance (5 positions along the gradient)
heat sum (5 positions along the gradient)
region (Jura, Central Plateau, Northern Prealps, Interior Valleys, Southern Prealps)

Theoretically, this results in 25 combinations per region.However, due to the partial correlation between heat sum and site water balance, not all of the possible combinations will be available.

NFI-plots belonging to each class of site water balance, heat sum and region will be used to evaluate the model predictions. Specifically, the model will be tested in its ability to predict accurately the dominant climax tree species at climatic equilibrium. Stand management and history data from the NFI will be used to select plots with an appropriate stand age to test such assumptions. Further, the model will be tested in its ability to predict the dominant seral species during succession. Again, management and stand age data will be used to select appropriate plots from the NFI database.

Proposed Work:
We will derive the model parameters as described above. The NFI will serve as a basis to select an appropriate number of plots to evaluate the models capability to simulate accurately stand composition, structure, and succession. Specifically, we will test the models ability to predict the:

climax dominant tree species (cover, species stand basal area, duration to appearance)
dominant seral tree species (cover, species stand basal area, duration of presence)
short term basal area increment between the two NFI sampling procedures
total stand leaf area at climax composition
total stand basal area at climax composition

Stand compositional predictions (climax and seral species) will be evaluated by tabulating simulated and actual (NFI) composition in a "confusion matrix", subject to a n x n c² goodness-of-fit test. We will predict total stand and species basal area, and 10-year basal area increments, compare the simulated results with actual BA data (from NFI), and calculate the RMS errors of predictions. We will use published regression coefficients to calculate leaf area index (LAI) from DBH or BA to estimate actual stand LAI for the selected NFI plots and compare these with simulated stand LAI data. Comparisons result in assessing the RMS error of predicted stand LAI.

4.4. Dispersal model

Goal:
Develop a seed dispersal module for this model in order to predict availability of seeds in space and time that potentially germinate at given locations in the simulated landscape.

Methods:
Plant migration is best viewed as a process that includes three basic elements; namely (1) the distance seeds are transported by a vector (Sauer 1988), (2) the time it takes for seeds to germinate, establish, grow and mature as trees, so that the next generation of seeds can be dispersed (Solomon & Kirilenko 1997), and (3) the amount of offspring produced per tree and surviving till maturity. The second and third part are dependent on biophysical and competitional parameters mostly covered by the forest simulator, while the first part is the primary aim of a seed dispersal module. Various studies have collected and analysed data on seed dispersal distances, particularly for North American plants (Brown et al. 1988; Chambers & McMahon 1994; Farmer 1997; Green 1980; Greene & Johnson 1989; 1995; Hutchins & Lanner 1982; Johnson & Adkisson 1985; Matlack 1987; Stapanian & Smith 1986; van der Wall & Balda 1977), to the most part for wind-dispersed seeds. To a lesser extent, such data is also available for Central European species abundant (Jensen 1985; Matlack 1987; Stoecklin & Baeumler 1996). Such studies most often list average seed dispersal as well as maximum observed dispersal distances (e.g. Cain et al. 1998).

However, when such maximum dispersal distances are re-scaled to migration rates per 100 years, the resulting rates are generally much slower (Cain et al. 1998; Greene & Johnson 1995; Skellam 1951; Webb 1986) than what has been observed from palynological records during the Holocene (e.g. Bennet & Lamb 1988; Birks 1989). The distinction between short- and long-distance dispersal is proposed to explain such discrepancies (Hengeveld 1989; Shigesada & Kawasaki 1997). Also, it seems reasonable to mathematically describe the two processes differently; namely (1) as a deterministic, sigmoidal or exponential function for short distance dispersal, with a decrease in seed mass with increasing distance, and (2) a stochastic long-distance dispersal (Fig. 2).

Fig. 2: The conceptual model of seed dispersal: Dispersal can be viewed as operating on to different scales, each on its own processes. See text for further explanations

We propose to formulate the long distance dispersal in a first version stochastically and to simulate the short-distance dispersal as a truncated exponential / sigmoidal function. The stochastic long-distance dispersal is limited by a maximum allowable distance, and by an inverse-distance weighted probability. In further model versions we will attempt to find an adequate deterministic approximation also for the long-distance dispersal.

Because information on seed production is scarce, we propose to keep the first model version as simple as possible by:

scaling seed availability between 1 (max) and 0 (min) (short-distance dispersal function).
adding up dispersed seeds from neighbouring pixels (as dispersed distance-dependent by the 2 processes).
assigning minimum rates for stochastic long-distance dispersal necessary to occasionally germinate.
ranking germination rates (5: very good; 4: good; 3: moderate; 2: poor; 1: very poor).
calculating regeneration rates as a function of locally simulated seed availability and germination rate (given that the biophysical environment is suitable).

Additionally, we will define DBH thresholds for individual tree species beyond which trees are assumed to be either too young (small) or too old to successfully produce viable seeds. The minimum DBH to reach maturity to produce seeds is an important element in calculating migration rates and lags in the readjustment of vegetation to changing environmental conditions (Solomon & Kirilenko 1997). We will use DBH rather than age, since trees in the shade of a well established overstory are clearly retarded in their development to maturity compared to trees that develop in full light. Long distance dispersal will be assessed by recorded maximum transport distances (Hughes et al. 1994) or properties of the seed-vectors, e.g. the radius of action of seed transporting birds (Johnson & Webb III 1989; Stimm & Böswald 1994).

Proposed Work:
We will develop a seed dispersal module that is computationally efficient, simple, and able to cope with the principle processes of seed dispersal. The short-distance dispersal is mathematically formulated as a deterministic, truncated function, parameterised primarily from literature data on observed dispersal ranges. Long-distance dispersal will be formulated in a first model version as a stochastic process, which includes inverse-distance weighted probability, and a maximum allowable distance. Long-distance dispersal maxima are derived from literature, data on dispersal ranges as well. We will further attempt to approximate stochastic dispersal by a deterministic formulation.

Besides omni-directional seed dispersal, we will foresee a directional seed transport implementation. This will allow also to take into account seed transport by rivers or by migrations of humans.

The model evaluation with palynological data (see below) is intended to test both dispersal processes. Palynological sample sites within close distance will reveal how reasonable the short-distance dispersal parameters are to simulate migration of trees. Contrarily, migrational lags between remote sample sites will reveal the importance of long-distance dispersal parameters that aim at accelerating the migration (given large enough areas of suitable terrain and suitable climatic conditions).

4.5. Climate input scenarios

Goal:
Generate climate maps for selected time windows in the Holocene. The development of prehistoric climate scenarios is ideally independent of pollen data.

Methods:
Compared to other countries, in Switzerland exists a wealth of climatic archives (i.e. in glaciers and in lake or mire sediments Lister et al. 1998). However, climate reconstruction involves many uncertainties. Consequently, often rather relative temperature changes (cooler/warmer) than absolute temperatures are available. Most of the climatic reconstruction is based on plants, particularly tree pollen, and thus not suitable as input for simulations which are to be compared with pollen data. However, in the sediments used for pollen analyses there are also remains of aquatic vegetation (Ammann et al. 1994) or zooplankton (Lotter et al., in prep.). Some key-sites where multi-proxy data exist can be used for temperature reconstruction, similar to dO¹⁸-data (Lister 1988, 1989). Precipitation reconstruction is more difficult; Guiot et al. (1993) calculated prehistoric precipitation maps in a spatial resolution of 1° based on a combination of pollen data with prehistoric lake levels as archived in lake sediments.

These local climatic data can possibly be combined with continental and global climatic reconstruction from alpine or polar ice cores.

Proposed Work:
For the proposed study we will analyse sources of past climatic data (time windows in the Late Glacial and Holocene, see table 1) in order to generate prehistoric climatic maps necessary for the spatial application of the model.

Temperature: We will attempt to reconstruct past climatic scenarios from a combination of climate proxies (table 1) available for many lowland and some high-elevation sites. Attention to non-climatic effects such as residence-time is important. Holocene temperature scenarios will thus only be available for individual locations. Hence it is most promising to calculate the difference for each reconstructed location to modern climatic maps, spatially and statistically interpolate the deviations, and generate climatic maps by adding the deviations to modern climatic maps.

Table 1: Climate proxies for independent climate reconstruction


Time window in uncalibrated radiocarbon years	Tree Genera involved	Climate Proxy (selection)	Sites (selection)	Literature (selection)

14-11.5 ka	Betula "alba"	d¹⁸O	Gerzensee	Eicher 1987
	Pinus "non-cembra"	waterplants	Leysin	Ammann et al. 1998
	Juniperus communis	insects	Lobsigenseee	Elias & Wilkinson 1983; Ammann 1989; Ammann et al. 1994

11.5-8.5 ka	Ulmus	d¹⁸O	Gerzensee	von Grafenstein, in prep.
	Quercus	zooplankton	Gerzensee	Lotter et al. in prep.
	(Corylus)	waterplants	Origlio/TI	Tinner et al. submitted
	Abies
	Fagus

6.5-3 ka	Abies	waterplants	Wallisellen	(Haas 1996)
	Fagus
	(Carpinus)

For most recent time window special attention will be paid to the assessment of human impact that may have favoured immigration through increased gap frequency after forest clearance.

Precipitation: Fewer subfossil data are available than for temperature reconstruction. For the Swiss Plateau and for the Jura mountains records of lake-level changes can be taken as reflecting the effective moisture (i.e. P-ETp) for the period before wide-spread human impact (Magny 1993; Richoz 1998). We will downscale the coarse-resolution historic precipitation maps by calculating the difference between these precipitation maps and actual precipitation maps that are aggregated to the same spatial resolution. The differences will then be added to the actual precipitation maps to derive historical maps of precipitation in higher spatial resolution.

4.6. Model output to pollen conversion

Goal:
Convert model output to pollen data for palaeoecological data-model comparison.

Methods:
The model will produce numbers of trees of each species in each DBH-class, which can be translated to species specific total basal area and by empirical functions to species biomass. Furthermore, the species-specific LAI will be available.

To make the model output comparable to palynological data, the output variables have to be converted to species specific pollen data. The quantitative conversion of fossil pollen records into past plant abundance has been a major issue for palaeobotanists for a long time (see e.g. Sugita 1994). Traditionally, conversion factors of Iversen (Faegri & Iversen 1975) and Andersen (1970) are used that relate pollen fractions to the "representation in the vegetation", which we interpret as total basal area. Pollen data are in most cases given as fractions, which makes them very sensitive even to small biases in the observations or in the conversion factors. The drawbacks of these representation factors, such as heterogeneity in pollen source areas, have been discussed (Birks & Gordon 1985). However, using the representation factors is the only way to obtain an estimate of the relation between pollen and some variable characterizing species abundance, such as biomass or basal area.

Proposed Work:
We will transform the species specific basal areas to pollen percent with the factors used in Lischke et al. (1998a), which are based on those of of Iversen (Faegri&Iversen 1975) and Andersen (1970). Besides percent pollen values, we will use wherever available and reliable, the absolute pollen amount, which is necessary to assess the first appearance of a certain species at a sample site and by this migration speed (see e.g. Bennet & Lamb 1988).

The first appearance (i.e. immigration) and expansion (i.e. population growth) of tree taxa will obtained from the following elements:

Pollen percentages will be used to assess the absolute (first appearance), the empirical (continuous curve), and the rational (rapidly increasing) limits of the pollen types
Pollen concentrations (grains per cm² ) and pollen influx (grains per cm²and year) will be analysed, where ever available
Macrofossil remains

4.7. Simulations of Holocene migration pattern and adjustment of dispersal parameters

Goal:
Simulate selected time windows for known past climate trajectories along main migration avenues of selected tree species (e.g. Fagus, Abies, Picea, Betula), and compare the test results with measured tree species composition from palynological records.

Methods:
Once the past climate mapping is performed, we will calculate two versions of model tests for historic tree migration simulations. The first test will be performed in a limited area along the Swiss Plateau, between a limited number of test sites. The second test will be performed on the entire territory of Switzerland.

The simple test aims at evaluating the general model performance along uninterrupted spatial gradients. The model should reveal reasonable compositional dynamics of tree species at the test locations, and it should be able to predict the migrational lags between the study sites accurately for the tree species involved. Since the terrain is relatively simply in structure, it is not easy to distinguish whether the two seed dispersal processes both operate reasonably well.

The second test aims at evaluating the model behaviour on a complex terrain. Specifically, the model should be able to generate migration rates for tree species similarly to those generated by isochrone mapping from pollen sample sites. This test enables to better distinguish between short- and long-distance dispersal parameters. Of special interest is the ability of the model, to simulate +/- accurately the migration of trees into valleys (e.g. Rhone Valley), or over vs. around mountain passes (e.g. Gotthard, Brünig).

The model tests will reveal whether the initial dispersal parameters are reasonable, or whether the have to be adjusted. We will primarily readjust the long-distance dispersal parameters, and the parameters for regeneration success under various habitat conditions. We will assume that such parameters are generally less accurate than the short-distance dispersal parameters (which in many cases are derived from experimental data).

Proposed Work:
We will generate two sets of palynological data in order to test the model’s ability to predict migration of trees under climate trajectories accurately. The test data consist of:

a topographically homogenous, limited area along the Swiss Plateau including a few palynological test sites.
a large area covering the entire Swiss territory and a complex topography including a series of pollen test sites.

The two tests will be used to evaluate different aspects of the simulations; namely (1) the compositional dynamics, (2) the duration of migrational lags and migration rates, (3) the mechanism of short- and long-distance dispersal parameters for individual tree species. Depending on the tests we will readjust the long-distance seed dispersal parameters.

5. Schedule and Objectives

Performance Goals:

We will start the project at June 1, 1999. The respective dates are set accordingly.


Objectives Year 1	Responsible	by
Generate first version of new hybrid model from LAASIM and DisCForM	Zimmermann, Lischke	`Dec. '99`
Select sites for past migration simulations from ALPADABA	Ammann, Lischke	`Dec. '99`
Calibrate literature-based parameters	Zimmermann	`Mar. '00`
Select appropriate evaluationplots from the Swiss NFI database	Zimmermann, Lischke	`Jun. '00`
Develop maps for site water balance and minimum temperature	Zimmermann	`Jun. '00`
Develop climate reconstruction for reference sites	Ammannn, Lischke	`Jun. '00`

Objectives Year 2	Responsible	by
Generate second version of hybrid model with aggregated local dispersal	Lischke, Zimmermann	`Aug. '00`
Parameterise the new hybrid model for field/map-derived parameters	Zimmermann	`Aug. '00`
Evaluate the model on selected plots of the Swiss NFI	Zimmermann, Roberts, Lischke	`Oct. '00`
Develop a seed dispersal module	Zimmermann, Lischke	`Mar. '01`
Develop maps for past climate from actual climate and reference data	Zimmermann	`Jun. '01`

Objectives Year 3	Responsible	by
Test tree migration model on few simple transects for past time windows	Zimmermann, Ammann	`Sep. '01`
Simulate large-scale tree migration on for past time windows	Zimmermann, Ammann	`Dec. '01`
Fine-tune the dispersal module (if necessary)	Zimmermann	`Feb. '02`

Management Plan:

Researcher	Responsibilities
H. Lischke WSL, CH	Research: Forest simulation and ecology; quarterly project meetings; select sites from ALPADABA for past-migration simulations; aggregation of hybrid model. *Management*: budget management; report to NSF; supervise N. Zimmermann in forest ecology.
B. Ammann Univ. Bern, CH	*Research: Palynology and climate reconstruction; select sites from Alpine Palynological Univ. Bern Database for past-migration simulations. Management*: supervise independent climate reconstruction for past migration studies.
D.W. Roberts Utah State, USA	*Research: Forest simulation and ecology; statistical analysis of model misclassification/model improvement. Management*: supervise N. Zimmermann in forest ecology.
N.E. Zimmermann WSL, CH	*Research: Forest simulation and ecology, climate mapping; development of a hybrid between LAASIM and DisCForM; parameterization of the new forest simulator; generation of the missing climate maps; development of Holocene climate maps for the past climate migration study; design a seed dispersal module; perform past climate model runs. Management*: collaborate in reports to NF; maintain Web-based resources.

Our core research team includes scientists in forest ecology/forest modelling (Lischke, Roberts, Zimmermann) and in paleo-ecology (Ammann). Dr. Zimmermann will be supervised by Lischke and Roberts for model development and model evaluation. In climate mapping he will seek collaboration with Dr. F. Kienast from FSL (Birmensdorf), with whom he has collaborated fruitfully on this subject in the past.

Lischke will schedule and moderate quarterly project meetings, at which team members will present results, solicit comments, and set goals. Collaboration with Roberts will be maintained through email, ftp, and Web-based exchange.

6. Significance of the proposed research

6.1. Objectives and Impact

Implement an environmentally sensitive framework for simulating tree species migration

Fields: Ecological Risk Assessment, Forestry, Socio-Economy

Impacts: The physiologically mechanistic, spatially explicit migration model will serve for future studies of tree migrations (and of associated risk assessments) and of forest dynamics:

in the Holocene, examining past tree migrations in entire Europe recorded in the European Pollen Database (EU-Project planned) to give insights in the processes involved in migration and vegetation adaptations to fast and slow climatic changes .
in the present, examining the risk of invasion and spread of non-native species, which is important for nature conservation.
in the future, under scenarios of climate change to assess the adaptation potential of vegetation. The simulator will allow to implement the effects of human influences such as landscape fragmentation and anthropogenic seed transport. It will thus be suitable for assessing the potential of tree species to reach new favorable habitats, or the risks of losing genetic variability, and of local extinction and of invasion of (non-native) tree species. This will facilitate decision-making and planning in Forest Resources. It will be useful particularly also for the assessment of the rate of tree migration in the boreal zone, and thus for the assessment of the C-binding potential.

Develop a physiologically mechanistic model to simulate the successional dynamics of forest stands

Fields: Ecophysiology, Forest Ecology, Population Biology

Impacts: The implementation of a physiologically mechanistic model is innovative in its own right. Recent advances in large-scale modelling of ecosystem photosynthesis and carbon cycling make it desirable to incorporate such approaches into a distribution based dynamic forest succession model. The incorporation of better physiological processes allows simulating biological interactions more mechanistically and, thus, more trustworthy, under non-equilibrium conditions.

Implement a forest succession model as spatially explicit

Fields: Biogeography, Forestry, Landscape Ecology

Impacts: Spatial autocorrelation and neighborhood interactions are of great importance for the understanding of ecological systems at the stand and the landscape level. So far, most of the existing forest succession models were implemented to run at given isolated locations. Alternatively, they were applied iteratively to sets of raster points in a landscape. A spatially explicit model , on the other hand, is capable of integrating spatial processes at variable dimensions and resolutions, and facilitates data exchange with existing Geographical Information Systems.

Implement a seed dispersal module to a forest succession model

Fields: Biogeography, Landscape Ecology, Population Biology

Impacts: Plant species are believed to behave and react individually to changes in environments and to differ individually in basic requirements of energy, water, CO2 and nutrients. The unique pattern of survival and re-colonization of formerly glaciated areas by different tree species is an expression of their individualistic behavior. Seeds are the principal driver of plant migration, and realistic pattern of seed availability to potentially germinate at given locations are crucial for the development of a patchy and heterogeneous landscape. The implementation of a seed dispersal module will give new insight into processes that generate heterogeneity at medium to large spatial scales.

Aggregate an individual based model to population-based model

Fields: Theoretical Ecology, Population Biology, Modeling Techniques

Impacts: The aggregation of the individual-based model to a population model will have to take into account the spatial heterogeneity resulting from local interactions and dispersal, and the temporal variability caused by fluctuating environmental conditions. The aggregation will give insights into the importance of individual dynamics to the entire populations’ dynamics, and into the emerging properties. Besides, the developed approaches can be used for upscaling of similar ecological models.

Apply the model in scenario studies to specific time-windows along past-migration pathways

Fields: Biogeography, Palaeo-Ecology

Impacts: Several studies have focused on Holocene isochrone maps, and patterns and rate of spread of tree species in Europe and England following the post-glacial warming. These empirical studies revealed valuable information on the rate of spread of specific tree species under specific pre-historical circumstances. Successful integration of a seed dispersal module and modelling of the dynamics of the component tree species will give insight into the mechanisms of these spatial and temporal patterns. We do not assume to accurately simulate the vegetation history of the Holocene. Still, simulating the re-colonization of trees along principal migrational pathways in scenario studies for selected time windows (and comparing the results with empirically derived rates of spread) will lead towards a better understanding of vegetation processes and evolution at the landscape level due to environmentalchange.

7. International collaboration

Besides our Swiss collaboration between WSL, ETH and the University of Bern we will mainly interact with the lab of David W. Roberts (Logan, UT), who developed the initial version of LAASIM. This model is currently in use in different projects related to simulate the forest dynamics and dependent processes like mountain pine beetle dynamics and spread. We, thus, will also maintain collaborations with J.A. Logan (USDA-Forest Service, Logan, UT) and J. Powell (Math. Dept., Logan, UT), whose expertise in the translation of stochastic processes into deterministic, and mathematical tractable formulations is highly welcome. Concerning the model aggregation, and the theoretical aspects of structured, stochastic population models, in particular in forest modelling, we collaborate with the group of Odo Diekmann in Utrecht (NL), with Joan Saldana in Barcelona (SP) and with the group of Pierre Auger in Lyon (FR). Related to the development of a physiology-based forest model, we are in contact with other forest modelers, particularly with the group of the Potsdam Institute for climate impact research (PIK).

Additionally, we collaborate with various research groups in Europe in the planning of a European Union-based project to simulate large-scale Holocene tree migrations over Europe. It is planned to use the same modelling tool we propose to develop in this project; a hybrid between the computationally efficient DisCForM and the physiologically mechanistic and spatially explicit LAASIM. This study will be an important test of several hypotheses on Holocene tree migration patterns and on Galcial refuges. The simulated patterns will be tested against pollen data from the European Pollen Database. Current collaborators are: J. Guiot (France) and A. Lotter and Felix Kienast (Switzerland).

8. References:

Ammann, B. 1989. Late-Quaternary palynology at Lobsigensee. Regional vegetation histroy and local lake development. Dissertationes Botanicae 137:157.

Ammann, B., M. Andree, L. Chaix, U. Eicher, S. Elias, W. Hofmann, H. Oeschger, U. Siegenthaler, K. Tobolski, B. Wilkinson, and H. Züllig. 1985. Lobsigensee - Late-Glacial and Holocene environments of a lake on the Central Swiss Plateau. Studies in the Late Quaternary of Lobsigensee 11. In Swiss Lake and Mire Environments During the Last 15 000 Years., edited by G. Lang.

Ammann, B. , J. Birks, S. Brooks, U. Eicher, U. von Grafenstein, W. Hofmann, W.O. van der Knaap, G. Lemdahl, J. Schwandewr, K. Tobolski, and L Wick. 1998. Wie reagieren aquatische und terrestrische Oekosysteme auf rasche Klimaänderungen?, Reports on NFP31. Zürich: Vdf Verlag.

Ammann, B., L. Chaix, U. Eicher, S.A. Elias, M-J. Gaillard, W. Hofmann, U. Siegenthaler, K. Tobolski, and B. Wilkinson. 1983. Vegetation, insects, molluscs and stable isotopes form late Würm deposits at Lobsigensee (Swiss Plateau). Studies in the late Quaternary of Lobsigensee 7. Rev. Paleobiologie 2 (2):163-180.

Ammann, B., A.F. Lotter, U. Eicher, M.J. Gaillard, B. Wohlfarth, W. Haeberli, G. Lister, M. Maisch, G. Niessen, and C. Schlüchter. 1994. The Würmian Late-glacial in lowland Switzerland. Journal of Quaternary Science 9 (2):119-125.

Ammann, B., and L. Wick. 1993. Analysis of fossil stomata of conifers as indicators of the alpine tree line fluctuations during the Holocene. In Oscillations of the Alpine and Polar Tree Limits in the Holocene, edited by B. e. a. Frenzel.

Andersen, S.T. 1970. The relative pollen productivity and pollen representativity of North European trees and correction factors for tree pollen spectra. Danm. geol. Unders. Ser. II 96:1-99.

Auger, P.M., and R. Roussarie. 1994. Complex ecological models with simple dynamics: From individuals to populations. Acta Biotheoretica 42:111-136.

Bazzaz, F.A. 1979. The physiological ecology of plant succession. Ann. Rev. Ecol. Syst. 10:351-371.

Bennet, K.D. 1983. Postglacial population expansion of forest trees in Norfolk, UK. Nature 303:164-167.

Bennet, K.D., and H.F. Lamb. 1988. Holocene Pollen Sequences as a record of competitive interactions among tree populations. Trends in Ecology and Evolution 3 (6):141-144.

Birks, H.J.B. 1989. Holocene isochrone maps and patterns of tree-spreading in the British Isles. Journal of Biogeography 16:503-540.

Birks, H.J.B. 1993. Quaternary paleoecology and vegetation science - current contributions and possible future developments. Reviews in Palaeobotany and Palynology 79:153-177.

Birks, H.J.B., Gordon, A.D. 1985. Numerical Methods in Quaternary Pollen Analysis. London: Academic Press.

Boardman, N.K. 1977. Comparative photosynthesis of sun and shade plants. Ann. Rev. Plant Physiol. 28:355-377.

Bolker, B., and S.W. Pacala. 1997. Using moment equations to understand stochastically driven spatial pattern formation in ecological systems. Theoretical Population Biology 52 (3):179-197.

Bonan, G.B., and L. Sirois. 1992. Air temperature, tree growth, and the northern and southern range limits to Picea mariana. Journal of Vegetation Science 3 (4):495-506.

Bossel, H. 1987. A dynamic simulation model of tree development under pollution stress. Paper read at
Erwin-Riesch-Workshop, Advances in Systems Analysis of Biological processes. 2nd Ebernburger Working Conference. Vieweg, Braunschweig.

Bossel, H. 1991. Modelling forest dynamics: Moving from description to explanation. For. Ecol. Manage. 42:129-142.

Bossel, H., H. Krieger, H. Schäfer, and N. Trost. 1991. Simulation of forest stand dynamics, using real-structure process models. Forest Ecology and Management 42:3-21.

Botkin, D.B., J.F. Janak, and J.R. Wallis. 1972a. Rationale, limitations and assumptions of a northeastern forest growth simulator. IBM Journal of Research and Development 16:101-116.

Botkin, D.B., J.F. Janak, and J.R. Wallis. 1972b. Some ecological consequences of a computer model of forest growth. Journal of Ecology 60:849-872.

Brix, H, and K. Mitchell. 1983. Thinning and nitrogen fertilization effects on sapwood development and relationships of foliage quantity to sapwood area and basal area in Douglas fir. Can. J. For: Res. 13:384-389.

Brown, K. R., D. B. Zobel, and J. C. Zasada. 1988. Seed dispersal, seedling emergence, and early survival of Larix laricina (DuRoi) K. Koch in the Tanana Valley, Alaska. Canadian Journal of Forest Research 18 (3):306-314.

Bugmann, H. 1994. On the ecology of mountainous forests in a changing climate: A simulation study. Diss. ETH Nr. 10638, Environmental Sciences, Swiss Federal Institute of Technology Zurich, Zurich.

Bugmann, H. K. M. 1996. A simplified forest model to study species composition along climate gradients. Ecology 77 (7):2055-2074.

Bugmann, H. 1997. Sensitivity of forests in the European Alps to future climatic change. Climate Research 8:35-44.

Bugmann, H., and A. Fischlin. 1996. Simulating forest dynamics in a complex topography using gridded climatic data. Climatic Change 34 (2):201-211.

Bugmann, H., R. Grote, P. Lasch, M. Lindner, and F. Suckow. 1997. A new forest gap model to study the effects of environmental change on forest structure and functioning. In Impacts of Global Change on Tree Physiology and Forest Ecosystems: proceedings of the international conference on impacts of global change on tree physiology and forest ecosystems, held 26-29 November 1996, Wageningen, the Netherlands, edited by G. M. J. Mohren, K. Kramer and S. Sabate. Dordrecht: Kluwer.

Cain, M. L., H. Damman, and A. Muir. 1998. Seed dispersal and the holocene migration of woodland herbs. Ecological Monographs 68 (3):325-347.

Catellino, P.J., I.R. Noble, R.O. Slatyer, and S.R. Kessel. 1979. Predicting the multiple pathways of plant succession. Journal of Environmental Management 3:41-50.

Chambers, J.C.Macmahon, J.A. 1994. A day in the life of a seed: movements and fates of seeds and their implications for natural and managed systems. Annu. Rev. Ecol. Syst. 25:263-292.

Cleve, K. van, and L. K. Oliver. 1982. Growth response of postfire quaking aspen (Populus tremuloides Michx.) to N, P and K fertilization. Canadian Journal of Forest Research 12 (2):160-165.

Davis, M.B. 1983. Holocene vegetational history of the eastern United States. In Late-Quaternary environments of the United States, edited by H. E. Wright. Minneapolis, Minnesota, USA: University of Minnesota.

Davis, M.B. 1989. Lags in vegetation response to greenhouse warming. Climatic Change 15:75-82.

Davis, M.B., K.D. Woods, S.L Webb, and R.P. Futyma. 1986. Dispersal versus Climate: Expansion of Fagus and Tsuga into the Upper Great Lakes. Vegetatio 67:93-103.

Deutschman, D. 1998. Emergent Dynamics in a Forest Model: The Role of Local Competition and Limited Dispersal in Mediating Broad-scale Patterns in SORTIE.

Deutschman, D.H., S.A. Levin, C. Devine, and L.A. Buttel. 1997. Scaling from Trees to Forests: Analysis of a Complex Simulation Model. Science online 277 (5332):ca. 37, cf. http://www.sciencemag.org/.

Dixon, R.K., S. Brown, R.A. Houghton, A.M. Solomon, M.C. Trexler, and J. Wisneiwski. 1994. Carbon pools and carbon flux of global forest ecosystems. Science 263:185-190.

Dyer, J.M. 1995. Assessment of climatic warming using a model of forest species migration. Ecol. Model. 79 (1-3):199-219.

EAFV. 1988. Schweizerisches Landesforstinventar. Ergebnisse der Erstaufnahme 1982-1986. Vol. 305, Ber.
Eidgenöss. Forsch.anst. Wald Schnee Landsch. Birmensdorf: Eidgenössische Anstalt für das forstliche Versuchswesen in Zusammenarbeit mit dem Bundesamt für Forstwesen und Landschaftsschutz.

Eicher, U., and U. Siegenthaler. 1983. Stable isotopes in lake marl and mollusc shells from Logsigensee (Swiss Plateau). Studies in the Late Quaternary of Lobsigensee 6. Revue de Paleobiologie 2:217-220.

EJPD-Bundesamt für Raumplanung, EVD-Bundesamt für Landwirtschaft, and EDI- Bundesamt für Forstwesen. 1980. Bodeneignungskarte der Schweiz. Grundlagen für die Raumplanung. Bern: Eidg. Drucksachen- und Materialzentrale.

Elias, S.A., and B. Wilkinson. 1983. Lateglacial insect fossil assemblages from Lobsigensee 3. Revue de Paleobiologie 2:189-204.

Elias, S.A., and B. Wilkinson. 1985. Fossil assemblages of Coleoptera and Trichoptera at Lobsigensee. In Swiss Lake and Mire Environments during the last 15 000 years, edited by G. Lang. Stuttgart: Cramer.

Faegri, K., and J. Iversen. 1975. Textbook of pollen analysis. Copenhagen, Denmark: Munksgaard.

Farmer, R.E. Jr. 1997. Seed ecophysiology of temperate and boreal zone forest trees. Translated by hl. Delray Beach: St. Lucie Press.

Finegan, B. 1984. Forest Succession. Nature 312:109-114.

Fischlin, A., H. Bugmann, and D. Gyalistras. 1995a. Sensitivity of a forest ecosystem model to climate parametrization schemes. Environmental Pollution 87 (3):267-282.

Fischlin, A., H. Lischke, and H. Bugmann. 1995b. The fate of forests in a changing climate: Model validation and simulation results from the Alps: Systems Ecology, Inst. of Terr. Ecology, ETH Zürich.

Fulton, M.R. 1991. A computationally efficient forest succession model: Design and initial tests. Forest Ecology and Management 42:23-34.

Gear, A.J., and B. Huntley. 1991. Rapid changes in the range limits of Scots Pine 4000 years ago. Science 251:544-547.

Gholz, H. L. 1982. Environmental limits on aboveground net primary production, leaf area, and biomass in vegetation zones of the Pacific Northwest. Ecology 63:469-481.

Grabherr, G., M. Gottfried, and H Pauli. 1994. Climate effects on mountain plants. Nature 369:448.

Green, D.S. 1980. The terminal velocity and dispersal of spinning samaras. American Journal of Botany 67:1218-1224.

Greene, D. F., and E. A. Johnson. 1989. A model of wind dispersal of winged or plumed seeds. Ecology, USA 70 (2):339-347.

Greene, D. F., and E. A. Johnson. 1995. Long-distance wind dispersal of tree seeds. Canadian Journal of Botany 73 (7):1036-1045.

Grier, C. C., and S. W. Running. 1977. Leaf area of mature northwestern coniferous forests: relation to site water balance. Ecology 58 (4):893-899.

Guiot, J., S.P. Harrison, and I.C. Prentice. 1993. Reconstruction of Holocene precipitation patterns in Europe using pollen and lake-level data. Quaternary Research 40:139-149.

Haas, J.N., I. Richoz, W. Tinner, and L. Wick. 1998. Synchronous Holocene climatic oscillations recorded on the Swiss Plateau and at timberline in the Alps. Holocene 8 (3):301-309.

Hengeveld, R. 1989. Dynamics of Biological Invasions. London, UK: Chapman and Hall.

Horn, H.S. 1975. Forest Succession. Scientific American 232:90-98.

Houghton, J.J., L.G. Meiro-Filho, B.a. Callander, N. Harris, A. Kattenberg, and K. Maksell. 1996. Climate change 1995 - The science of climate change: Contribution of working group I to the second assessment report of the intergovernmental panel on climate change. .

Hughes, L., M. Dunlop, K. French, M.R. Leishman, B. Rice, L. Rodgerson, and M. Westoby. 1994. Predicting dispersal spectra: a minimal set of hypotheses based on plant attributes. In : Ecology and Biogeography of Mediterranean Ecosystems in Chile, California and Australia. Series-Title: Ecological Studies : Analysis and Synthesis. Springer, New York, edited by M. T. K. Arroyo, P. H. Zedler and M. D. Fox.

Hutchins, H.E., and R.M. Lanner. 1982. The central role of Clark's nutcracker in the dispersal and establishment of whitebark pine. Oecologia 55:192-201.

Iwasa, Y., V. Andreasen, and S.A. Levin. 1987. Aggregation in model ecosystems. I. Perfect aggregation. Ecological Modelling 47:287-302.

Iwasa, Y., S.A. Levin, and V. Andreasen. 1989. Aggregation in model ecosystems. II. Approximate aggregation. IMA Journal of Mathematics Applied in Medicine & Biology 6:1-23.

Jensen, T. S. 1985. Seed-seed predator interactions of European beech, Fagus silvatica, and forest rodents. Oikos 44:149-156.

Johnson, W.C., and C.S. Adkisson. 1985. Dispersal of beech nuts by blue jays in fragmented landscapes. American Midland Naturalist 113:319-324.

Johnson, W.C., and T. Webb III. 1989. The role of blue jays (Cyanocitta cristata L.) in the postglacial dispersal of fagaceous trees in eastern North America. J. Biogeogr. 16:561-571.

Karev, G.P. 1994. Structural models for natural forest dynamics. Doklady biological sciences: proceedings of the Academy of Sciences of the USSR 337:354-356.

Kienast, F. 1987. FORECE - A forest succession model for southern Central Europe. Oak Ridge, TN: Oakridge National Laboratories. Environmental Division.

Kirkilonis, M. 1998. Modelling Forests: On Stochastic Processes Modelling Tree Distributions.

Kirschbaum, M., and A. Fischlin. 1996. Climate change impacts on forests. In Climate change 1995 - Impacts, adaptations and mitigation of climate change: Scientific-technical analyses: Contribution of working group II to the second assessment report of the intergovernmental panel on climate change, edited by R. T. Watson, M. C. Zinyowera and R. H. Moss. Cambridge: Cambridge University Press.

Kloetzli, F, Walther, G.R., G. Carraro, and A. Grundmann. 1996. Anlaufender Biomwandel in Insubrien. Verhandlungen der Gesellschaft für Ökologie (Bremen) 26:537-550.

Kohyama, T. 1989. Simulation of the structural development of warm-temperate rain forest stands. Annals of Botany 63:625-634.

Kohyama, T. 1992. Size-structured multi-species model of rain forest trees. Funct. Ecol. 6 (2):206-212.

Kohyama, T. 1993. Size-structured tree populations in gap-dynamic forest: The forest architecture hypothesis for the stable coexistence of species. J. Ecol. 84 (2):207-218.

Kohyama, T., and T. Hara. 1989. Frequency distribution of tree growth rate in natural forest stands. Ann. Bot. 64 (1):47-58.

Kraeuchi, N., and F. Kienast. 1993. Modelling subalpine forest dynamics as influenced by a changing environment. Water Air and Soil Pollution 68 (1-2):185-197.

Kraev, Y. 1998. A Partial Differential Equation Modeling Forest Growth. Master thesis, Mathematics, ETHZ, Zürich.

Kullman, L., and O. Engelmark. 1991. Historical biogeography of Picea abies (L.) Karst. at its subarctic limit in northern Sweden. Journal of Biogeography 18:63-70.

Leemans, R., and I.C. Prentice. 1989. FORSKA, a general forest succession model: Institute of Ecological Botany.

Lischke, H. 1998. Veränderungen der Artenzusammensetzung der Schweizer Wälder bei einem schnellen Klimawechsel: Simulationsstudien. Bulletin: Kompetenzzentrum Holz, (in press).

Lischke, H., A. Fischlin, and A. Lotter. 1998a. Untangling a Holocene pollen record with forest model simulations and independent climate data. Ecology submitted.

Lischke, H., A. Guisan, A. Fischlin, J. Williams, and H. Bugmann. 1998b. Vegetation responses to climate change in the Alps - Modeling studies. In A view from the Alps: Regional perspectives on climate change, edited by P. Cebon, U. Dahinden, H. Davies, D. Imboden and C. Jaeger. Boston: MIT Press.

Lischke, H., T.J. Löffler, and A. Fischlin. 1998c. Aggregation of individual trees and patches in forest succession models - Capturing variability with height structured random dispersions. Theoretical Population Biology:(in press).

Lister, G., D.M. Livingstone, B. Ammann, D. Ariztegui, W. Haeberli, A.F. Lotter, C. Ohlendorf, C. Pfister, J.
Schwander, F. Schweingruber, B. Stauffer, and Michael Sturm. 1998. Alpine Palaeoclimatology. In A view from the Alps: Regional perspectives on climate change, edited by P. Cebon, U. Dahinden, H. Davies, D. Imboden and C. Jaeger. Boston: MIT Press.

Lister, G.S. 1988. A 15,000-Year isotopic record from Lake Zürich of deglaciation and climatic change in Switzerland. Quaternary Research 29:129-141.

Lister, G.S. 1989. Reconstruction of palaeo air temperature changes from oxygen records in Lake Zürich: The significance of seasonality. Ecologiae geologicae Helveticae 82 (1):219-234.

Loehle, C., and D. LeBlanc. 1996. Model-based assessments of climate change effects on forests: A critical review. Ecological Modelling 90 (1):1-31.

Löffler, T., and H. Lischke. 1999. Influences of Variabilities in Aggregating a Forest Model. Ecological Modelling:(in prep.).

Lotter, A.F. 1989. Evidence of annual layering in Holocene sediments of Soppensee, Switzerland. Aquatic Sciences 51 (1).

Lotter, A.F., H.J.B. Birks, W. Hofmann, and A. Marchetto. 1997. Modern diatom, cladocera, chironomid, and chrysophyte cyst assemblages as quantitative indicators for the reconstruction of past environmental conditions in the Alps .1. Climate. Journal of Paleolimnology 18 (4):395-420.

Lotter, A. F., H. J. B. Birks, W. Hofmann, and A. Marchetto. 1998. Modern diatom, cladocera, chironomid, and chrysophyte cyst assemblages as quantitative indicators for the reconstruction of past environmental conditions in the Alps. II. Nutrients. Journal of Paleolimnology 19 (4):443-463.

Lotter, A.F., and F. Kienast. 1992. Validation of a forest succession model by means of annually laminated sediments. Paper read at INQUA workshop on laminated sediments, at Lammi, Finland.

Magny, M. 1993. Holocene fluctuations o lake levels in the French Jura and sub-Alpine ranges, and their implications for past general circulation patterns. The Holocene 3 (4):306-313.

Malanson, G. P., and M. P. Armstrong. 1996. Dispersal probability and forest diversity in a fragmented landscape. Ecological Modelling 87 (1-3):91-102.

Margolis, Hank, Ram Oren, David Whitehead, and Merill R. Kaufmann. 1995. Leaf Area Dynamics of Conifer
Forests. In Ecophysiology of Coniferous Forests, edited by W. K. Smith and T. M. Hinckley. San Diego: Academic Press.

Matlack, G. R. 1987. Diaspore size, shape, and fall behavior in wind-dispersed plant species. American Journal of Botany 74 (8):1150-1160.

Mladenoff, David J., George E. Host, Joel Boeder, and Thomas R. Crow. 1993. LANDIS: A spatial model of forest landscape disturbance, succession, and management. In Environmental Modeling with GIS, edited by M. F.
Goodchild, B. O. Parks and L. T. Steyaert. New York: Oxford University Press.

Moore, A.D. 1990. The semi-Markov process: A useful tool in the analysis of vegetation dynamics for management. Journal of Environmental Management 30:111-130.

Noble, I.R., and R.O. Slatyer. 1977. Post-fire succession of plants in Mediterranean ecosystems. Paper read at Symposium on the Environmental Consequences of Fire and Fuel Management in Mediterranean Climate Ecosystems.

Noble, I.R., and R.O. Slatyer. 1980. The use of vital attributes to predict successional changes in plant communities subject to recurrent disturbance. Vegetatio 43:5-21.

Pacala, S. W., C. D. Canham, J. Saponara, J. A. Silander, R. K. Kobe, and E. Ribbens. 1996. Forest Models Defined By Field Measurements: Estimation, Error Analysis and Dynamics. Ecological Monographs 66 (1):1-43.

Pacala, S. W., C. D. Canham, and J. A. Silander, Jr. 1993. Forest models defined by field measurements: I. The design of a northeastern forest simulator. Canadian Journal of Forest Research 23 (10):1980-1988.

Pacala, S.W., and D.H. Deutschman. 1995. Details that matter: the spatial distribution of individual trees maintains forest ecosystem function. OIKOS 74:357-365.

Pastor, J., and W.M. Post. 1986. Influence of climate, soil moisture, and succession on forest carbon and nitrogen cycles. Biogeochemistry 2:3-27.

Portnoy, S., and M.F: Willson. 1993. Seed dispersal curves: behavior of the tail of the distribution. Evolutionary Ecology 7:25-44.

Prentice, I.C. 1986. Vegetation response to past climatic variation. Vegetatio 67:131-141.

Prentice, I. C., W. Cramer, S. P. Harrison, R. Leemans, R. A. Monserud & A. M. Solomon. 1992. A global biome model based on plant physiology and dominance, soil properties and climate. J. Biogeogr. 19: 117-134.

Prentice, I. C., M. T. Sykes, and W. Cramer. 1993. A simulation model for the transient effects of climate change on forest landscapes. Ecological Modelling 65 (1-2):51-70.

Pretzsch, H. 1995. Zum Einfluss des Baumverteilungsmusters auf den Bestandeszuwachs (On the effect of the spatial distribution of trees on the stand growth). Allgemeine Forst- und Jagd-Zeitung 166 (9-10):190-201.

Pretzsch, H. 1992. Konzeption und Konstruktion von Wuchsmodellen für Rein- und Mischbestände: Forstwissenschaftliche Fakultät der Universtität München und der bayer. forstlichen Versuchs- und Forschungsanstalt.

Richoz, I. 1998. Etude paléoécologique du lac de Seedorf (Fribourg, Suisse). Dissertationes Botanicae 293:177 pp.

Ritchie, J.C. 1986. Climate change and vegetation response. Vegetatio 67:65-74.

Roberts, D. W. 1996a. Landscape vegetation modelling with vital attributes and fuzzy systems theory. Ecological Modelling 90 (2):175-184.

Roberts, D. W. 1996b. Modelling forest dynamics with vital attributes and fuzzy systems theory. Ecological Modelling 90 (2):161-173.

Roberts, D. W., R.F. Fisher, J.M. Long, and S.N. Jack. 1993. The Leaf Area Allocation Model. In Simulation of Rocky Mountain Forest Dynamics and Climate Change.

Roberts, L. 1989. How fast can trees migrate? Science 243:735-737.

Running, S. W. 1994. Testing FOREST-BGC ecosystem process simulations across a climatic gradient in Oregon. Ecological Applications 4 (2):238-247.

Running, S. W., and J. C. Coughlan. 1988. A general model of forest ecosystem processes for regional applications: I. Hydrologic balance, canopy gas exchange, and primary production processes. Ecological Modelling 42:125-154.

Running, S. W., and S.T. Gower. 1991. FOREST-BGC, A general model of forest ecosystem processes for regional applications: II. Dynamic carbon allocation and nitrogen budgets. Tree Physiology 9:147-160.

Saldana, J., Kirkilonis, M. 1998. Modelling Forests: Height Structured Trees.

Santer, B.D., T.M.L. Wigley, T.P. Barnet, and E. Anyamba. 1996. Detection of climate change and attribution of cause. In Climate change 1995: the science of climate change, edited by J. T. Houghton, L. G. Meira Filho, B. A.
Callander, N. Harris, N. Kattenberg and K. Maskell. Cambridge, UK: Cambridge University Press.

Sauer, Jonathan D. 1988. Plant Migration. The dynamics of geographic patterning in seed plant species. Berkley, CA: University of California Press.

Shigesada, N., and K. Kawasaki. 1997. Biological Invasions: Theory and Practice. Edited by R. M. May and P. H. Harvey, Oxford Series in Ecology and Evolution. Oxford, UK: Oxford University Press.

Shugart, H. H. 1984. A Theory of Forest Dynamics: The Ecological Implications of Forest Succession Models. New York: Springer-Verlag.

Skellam, J.G. 1951. Random dispersal in theoretical populations. Biometrika 38:196-218.
Smith, T. M., and M. A. Huston. 1989. A theory of the spatial and temporal dynamics of plant communities.
Vegetatio 83:49-69.

Smith, T.M., and D.L. Urban. 1988. Scale and resolution of forest structural pattern. Vegetatio 74:143-150.
Solomon, A. M., and P.J. Bartlein. 1992. Past and future climate change: response by mixed deciduous-coniferous forest ecosystems in northern Michigan. Canadian Journal of Forest Research 22:1727-1738.

Solomon, A. M., and A. P. Kirilenko. 1997. Climate change and terrestrial biomass: what if trees do not migrate! Global Ecology and Biogeography Letters 6 (2):139-148.

Solomon, A.M., D.C. West, and J.A Solomon, eds. 1981. Simulating the role of climate change and species immigration in forest succession. Edited by D. C. West, H. H. Shugart and D. B. Botkin, Forest succession:
Concepts and application. New York, Heidelberg, Berlin: Springer.

Stapanian, M. A., and C.C. Smith. 1986. How fox squirrels influence the invasion of prairies by nut-bearing trees. Journal of Mammology 67:326-332.

Stimm, B., and K. Böswald. 1994. Die Häher im Visier. Zur Ökologie und waldbaulichen Bedeutung der Samenausbreitung durch Vögel. Forstwissenschaftliches Centralblatt 113:204-223.

Stoecklin, J., and E. Baeumler. 1996. Seed rain, seedling establishment and clonal growth strategies on a glacier foreland. Journal of Vegetation Science 7:45-56.

Sugita, S. 1994. Pollen representation of vegetation in quaternary sediments: theory and method in patchy vegetation. J. Ecol. 82 (4):881-897.

Tinner, W., B. Ammann, and P. Germann. 1996. Treeline fluctuations recorded for 12 500 years by soil profiles, pollen, and plant macrofossils in the Central Swiss Alps. Arctic and Alpine Research 28 (2):131-147.

Urban, D.L., and H.H. Shugart. 1989. Forest response to climate change: A simujlation study for southeastern forests. In The Potential Effects of Global Climate Change on the United States, edited by J. B. Smith and D. Tirpak. Washington: U.S. EPA.

van der Knaap, W.O., and B. Ammann. 1997. Depth-age relationships of 25 well-dated Swiss Holocene pollen sequences archived in the Alpine Palynological Data Base. Revue de Paléobiologie 16: 433-480.

van der Wall, S.B., and R.P. Balda. 1977. Coadaptations of the Clark's nutcracker and the pinon pine for efficient seed harverst and dispersal. Ecological Monographs 47:89-111.

Waring, R.H., H.L. Gholz, C.C Grier, and M.L. Plummer. 1977. Evaluating stem conducting tissue as an estimator of leaf area in four woody angiosperms. Can.J.Bot. 55:1474-1477.

Watson, R.T., M.C. Zinyowera, and R. H. Moss. 1996. Climate change 1995 - Impacts, adaptations and mitigation of climate change: Scientific-technical analyses: Contribution of working group II to the second assessment report of the intergovernmental panel on climate change. Cambridge: Cambridge University Press.

Webb, S.L. 1986. Potential role of passanger pigeons and other vertebrates in the rapid Holocene migrations of nut trees. Quaternary Research 26:367-375.

Webb, T., III. 1981. The past 11 000 years of vegetational change in eastern North America. BioScience 31:501-506.

Welten, M. 1982. Vegetationsgeschichtliche Untersuchungen in den westlichen Schweizer Alpen: Bern-Wallis. Denkschriften Schweizerische Naturforschende Gesellschaft 95:1-104.

Wick, L. 1994. Early-holocene reforestation and vegetation change at a lake near the Alpine forest limit: Lago Basso (2250 m asl), Northern Italy. Diss. Bot. (Festschrift G. Lang) 234:555-563.

Wick, L. 1996. Spät- und postglaziale Vegetationsgeschichte in den Südalpen zwischen Comersee und Splügenpass (Norditalien). phD Thesis, Geobotany, University of Bern, Bern.

Wick, L., and W. Tinner. 1997. Vegetation changes and timberline fluctuations in the central airs as indicators of Holocene climatic oscillations. Arctic and Alpine Research 29 (4):445-458.

Will-Wolf, S., and D.W. Roberts. 1993. Fire and succession in oak-mapple upland forests: a modeling approach based on vital attributes. In John T. Curtis: 50 years of Wisconsin Plant Ecology, edited by J. S. Fralish, R. P.
McIntosh and O. L. Loucks. Madison, WI: Wisconsin Academy of Science, Arts and Letters.
Williamson, M. 1996. Biological Invasions. London, UK: Chapman & Hall.

WSL. 1998. Schweizerisches Landesforstinventar LFI. Datenbankauszug der
Erhebungen (1983-85 und) 1993-95. Birmensdorf: Eidg. Forschungsanstalt für Wald, Schnee und Landschaft (WSL).

Zimmermann, N.E. 1996. Ein klimasensitives, räumliches Vegetationsmodell für die alpine Stufe der Schweiz. Doctoral Thesis, Universität Bern, Bern.

Zimmermann, N.E., and F. Kienast. 1995. Das Klima lässt sich nicht kartieren - Klimakarten werden gerechnet. Informationsblatt des Forschungsbereiches Landschaftsökologie, WSL 27:1-4.

Zimmermann, N.E., and F. Kienast. in press. Predictive mapping of alpine grasslands in Switzerland: species versus community approach. Journal of Vegetation Science.

Zimmermann, N.E., and D.W. Roberts. in prep. Modelling spatially-explicit forest dynamics in the Central Rocky Mountains: leaf area, biomass, and site water balance at the tree, stand and landscape level. .

[top] [back] [home]

Last Updated on 4/19/99
By Niklaus E. Zimmermann