Search

A model is a simplification of our understanding of reality, but how much simplification is too much? We assessed model sensitivity for a new spatially-explicit individual-based gap model (IBM) SIBBORK to determine how spatial resolution (i.e., plot size) and explicit representation of light and space affect the simulated vegetation characteristics (e.g., structure and composition). The 3-dimensional (3-D) representation of light is the most computationally-expensive part of the simulation, so it is worthwhile to evaluate whether the time, resource and effort of spatial explicity improve predictive capabilities. In the simulation, gap dynamics occur at the plot-level, wherein a certain level of spatial homogeneity is assumed. Therefore, for the simulation of each ecosystem it is imperative to select the appropriate scale at which gap dynamics are simulated, i.e. the plot size. We conducted these sensitivity analyses and qualitatively compared model output to the descriptions of forest structure and composition at two different sites in central and southern Siberia. We systematically varied the plot size (100–900 m2) and found that a 100 m2 plot is most appropriate for the simulation of the central Siberian boreal forest ecosystem. Moreover, we found that the forest structure and composition simulated under the light environment generated using a 3-dimensional light ray tracing subroutine more closely resembled local and regional forest characteristics at both sites, whereas the growth and stand processes computed based on the simplified (1-D) light conditions approximated only large-scale (continental) average forest structure and biomass for the Eurasian boreal forest.

Forests produce and emit abundant non-methane volatile hydrocarbon species (VOCs) influencing the atmosphere chemistry and climate. Over a half century of research has produced significant understandings of the biochemistry and eco-physiology of biogenic VOCs. However, VOCs production is highly species-specific, and the impact of changes in species composition and abundance on VOCs emissions is unobservable in the time scales usually seen in field experiments. Prior modelling efforts are based on cases with an aggregate representation of vegetation. Individual-based models of forests simulate the dynamics of complex forest ecosystems based on birth, growth, and death of the individual trees comprising a simulated forest. Here an individual-based forest VOCs emissions model—UVAFME-VOC v1.0—is developed from the state-of-the-art individual-based forest gap model, UVAFME, coupled with a canopy VOCs emission model. UVAFME-VOC v1.0 implementation for the temperate deciduous forest in the southeastern United States is tested by comparisons to independent data in the region. A model application tested the hypothesis that the historical collapse of American Chestnut ( Castanea dentate ) resulted in the dominance of oak trees ( Quercus spp.) and enhanced isoprene emissions. The model demonstrated a capability to simulate the forest compositional and structural dynamics and forest isoprene emission dynamics. The simulations show isoprene emissions depend heavily on forest successional stage and species composition, suggesting that environmental change can affect forest VOCs emissions primarily by influencing forest species composition. Prediction of isoprene emissions, of other phytogenic volatile organic compounds, and of impacts on atmospheric chemistry of various global change agents (e.g., warming, rising CO 2 , ozone elevation, and herbivory) should explicitly consider forest diversity change. This individual-based model could provide widespread applications in addressing these problems.

Testing ecological models involves using independent data on model performance, which can be difficult or practically impossible to obtain. Individual based models of forest dynamics, or gap models, simulate the change of forests by computing the annual growth, birth and death of each tree at a location in a forest. The models are relatively simple and simulate results that can be translated to multiple response scales: Individual plant growth, population birth-death processes, stand environmental dynamics ( e.g ., evapotranspiration, element cycling, heat flux, etc .), landscape processes, and regional and global change. This paper reviews some of the approaches applied to testing gap models. Then, it demonstrates the testing of the performance of an individual-based gap model of forest dynamics, FAREAST, through comparison against independent data from China and across Russia. As part of this model testing, biomass simulation output for 93 locations is compared to independent field-collected inventory-data from 44 Russian forests, which span a broad range of forest types across Russia. FAREAST captures biomass dynamics and stabilization at specific locations bracketing the measured values. At Changbai Mountain, the model accurately predicts the community dynamics of complex mixed forest types present along an elevational gradient, as well as the broad regional compositional patterns across China and Russia. Validation of regional detailed landscape dynamics shows the model performs with fidelity with an average R 2 value of 0.74 for 87 comparisons and an average root mean square error of 10.8 tC ha −1 . Performance of the model for historical conditions implies the model's applicability across a broad region and suggests the usefulness of a detailed model for evaluating forest change to management and changing climate.

The boreal zone of Alaska is dominated by interactions between disturbances, vegetation, and soils. These interactions are likely to change in the future through increasing permafrost thaw, more frequent and intense wildfires, and vegetation change from drought and competition. We utilize an individual tree-based vegetation model, the University of Virginia Forest Model Enhanced (UVAFME), to estimate current and future forest conditions across sites within interior Alaska. We updated UVAFME for application within interior Alaska, including improved simulation of permafrost dynamics, litter decay, nutrient dynamics, fire mortality, and post-fire regrowth. Following these updates, UVAFME output on species-specific biomass and stem density was comparable to inventory measurements at various forest types within interior Alaska. We then simulated forest response to climate change at specific inventory locations and across the Tanana Valley River Basin on a 2 × 2 km2 grid. We derived projected temperature and precipitation from a five-model average taken from the CMIP5 archive under the RCP 4.5 and 8.5 scenarios. Results suggest that climate change and the concomitant impacts on wildfire and permafrost dynamics will result in overall decreases in biomass (particularly for spruce (Picea spp.)) within the interior Tanana Valley, despite increases in quaking aspen (Populus tremuloides) biomass, and a resulting shift towards higher deciduous fraction. Simulation results also predict increases in biomass at cold, wet locations and at high elevations, and decreases in biomass in dry locations, under both moderate (RCP 4.5) and extreme (RCP 8.5) climate change scenarios. These simulations demonstrate that a highly detailed, species interactive model can be used across a large region within Alaska to investigate interactions between vegetation, climate, wildfire, and permafrost. The vegetation changes predicted here have the capacity to feed back to broader scale climate-forest interactions in the North American boreal forest, a region which contributes significantly to the global carbon and energy budgets.

a b s t r a c t Because of human pressures, the need to understand and predict the long-term dynamics and develop- ment of subtropical dry forests is urgent. Through modifications to the ZELIG simulation model, including the development of species- and site-specific parameters and internal modifications, the capability to model and predict forest change within the 4500-ha Guanica State Forest in Puerto Rico can now be accomplished. Published datasets and additional data from the U.S. Forest Service Forest Inventory Anal- ysis were used to parameterize the new gap model, ZELIG-TROP. We used data from permanent plots

A linkage between the two major modeling approaches to forest dynamics, transition Markovian models and jabowa-foret type simulators, is generated by developing a compact model of forest dynamics. This patch transition model utilizes functional roles instead of species. The roles or types are based on the regeneration and mortality characteristics of tree species; specifically, the requirements of canopy gaps for regeneration and the capacity to create canopy gaps upon death. A gap-size plot can be assigned to each of a set of states defined according to dominance of one of the roles. Transition probabilities among these states and mean holding times in each transition lead to semi-Markovian analytical calculations of the stationary state probabilities. Forest dynamics, as the proportions of total canopy space occupied by each role in a collection of gap-size plots, can be analyzed and simulated using a chain of first-order differential equations to emulate the distributed time-delays. Additional fixed time-delays in the transition of every pair of states is also included to account for long latencies. In addition to simplifying the simulations, the resulting model can also utilize available results of the theory of semi-Markov processes; and therefore, can provide analytical guidance to the simulations, the feasibility of direct exploration of hypothesis and the possibility of fast computation from closed-form solutions and formulae. These advantages can especially be useful in the simulation of landscape dynamics and species-rich tropical forests.

Shrub ( Myrica cerifera ) thickets are a major vegetation type on barrier islands. To evaluate the effects of moisture stress on shrub thicket distribution, a water relation model, MCHOG, was developed for Hog Island, a barrier island along the eastern shore of Virginia, USA. The model consists of three submodels: a transpiration submodel based on the Penman-Monteith equation, a rooting-zone water-balance submodel, and water-table dynamics submodel using Darcy's law. Precipitation, relative humidity, wind speed, solar radiation, and air temperature data collected on Hog Island in 1991 were used with an hourly time step to simulate transpiration patterns for shrub thickets along a cross-island transect. Transpiration under sufficient soil moisture was defined as potential transpiration. Comparisons were made between the potential transpiration pattern and the transpiration patterns under various environmental change scenarios. Simulation results suggest that the shrub thicket transpiration pattern is sensitive to climate change, landscape change, and sea-level rise. Soil water limits transpiration of all shrub thickets on barrier islands. Elevationally related capillary rise from groundwater also plays an important role in determining shrub distribution patterns. Competition may exist for groundwater resources among individuals of shrub thickets and between shrubs and associated grasses.

We describe an approach to investigating and understanding the interactions between vegetation structure and ecosystem processes that uses simulation models as a framework for comparison and synthesis across ecosystems arrayed along environmental gradients. The models are individual-based vegetation simulators and compartment models of nutrient cycling and soil water relations. Applications focus on interactions and feedbacks between vegetation structure (species composition, size structure) and ecosystem processes (water balance, nutrient cycling), and how these relationships vary across environmental gradients. Preliminary results indicate that life-history traits of plants have a profound influence on system-level behaviors, and that differences between grasslands and forests can be attributed largely to contrasting traits of grasses and trees. Experiments with linked vegetation-ecosystem process models diverge from simulations with either model run independently, suggesting the importance of feedbacks between details of vegetation pattern and ecosystem processes. The development of a fully coupled vegetation-ecosystem process model that is sufficiently general to simulate systems dominated by multiple lifeforms presents several conceptual, logistical, and scaling challenges, but also provides for new opportunities in ecosystem theory.

The Forest Ecosystem Dynamics (FED) project involves the development of an integrated mathematical model which links individual submodels of soil processes, forest growth and succession, and radiative transfer. The model will accommodate spatial scales from local to regional, and temporal scales from physiological to long term ecological processes. In its integrated form, the model is designed to simulate a forest ecosystem operating in a “process-response” manner whereby natural or anthropogenically induced changes in the environment will elicit responses within the soil, vegetation, and radiation regime of a forest that will impact and modify each other. In order to insure maximum flexibility for both the modeler and the user, an “object-oriented” structure will be implemented. In this way, the model will provide a tool with which patterns and processes within northern forests resulting from global change can be predicted.

Tree growth models project the growth and development of forest ecosystems by increasing the size of each simulated tree in the forest on an annual or greater periodic basis. These models are often referred to as ‘tree models’ because they are based on the birth, growth and death of individual trees, and the characteristics of individual trees are aggregated to describe the stand. This paper distinguishes between two types of models based upon their major purpose for development and data requirements. Forest growth models are defined as those used to assess the yield of a managed forest under prescribed conditions and usually require large calibration data sets. Community dynamics models are defined as those applied to ecological questions about the feedback of environment and species characteristics on growth and yield, and have species-specific rather than site-specific data requirements. The structure and data requirements for each model type affects their regional and temporal applicability as well as their predictive value. A combination approach using the two types of models may be t the most general model of tree growth and stand development.

The concepts of model validation and model realism are reconsidered in terms of the domain of applicability of a model. Ambiguity in both the application and the conceptualization of model validation is a problem for clearly assessing the usefulness of models. Model desirability (as well as the concepts of model adequacy and model reliability) is developed in a set theoretic context which is intended to reduce the ambiguity that often obstructs clear discussions of model validation. A set of criteria for defining desirable models is outlined and a list of principles for increasing model desirability is provided. It is shown that models of known adequacy can be used to reduce sample size in traditional statistical sampling procedures. A derivation using models and Bayes' theorem is provided as an example of such an application.

The use of spectral analysis to elucidate the cyclic behavior in time series generated by a forest stand growth simulation model is discussed. A stand-level simulator, FORET, for an Appalachian deciduous forest is described. An estimate of the power spectral density of the total biomass time series is calculated. The power spectral density estimate indicates a dominant cyclic behavior with a period of about 200 years. In addition the spectral density is approximately bandlimited. This characteristic makes possible the application of the sampling theorem for analysis of sampling rates.

The succession of a mixed species stand of the Acadian Forest was simulated by adopting an approach taken by Botkin et al. (1972) (JABOWA) and later modified by Shugart and West (1977) (FORET). The model simulates the dynamic aspects of successional behaviour of the stand by projecting the current stage of vegetation composition (size of each individual of each species present on a 1/12 ha plot) with an element of randomness. The growth of each tree is modeled as a function of its size (represented by its diameter at breast height (dbh), height and leaf area index), the climate, the tree's tolerance to shade and the competition for available resources and space. Stand aboveground biomass (metric ton/ha), leaf area index, number of trees/ha and species composition of the stand were simulated under seven different conditions: (a) normal treatment, (b) increasing biomass maximum 30%, (c) decreasing biomass maximum 30%, (d) increasing degree-days 30%, (e) decreasing degree-days 30%, (f) increasing light extinction coefficient 30% and (g) decreasing light extinction coefficient 30%. Under the first set of conditions (normal treatment), the aboveground biomass reached a maximum value during the first hundred years, decreased during the second, and remained stable for the rest of the 500-year simulation period. The leaf area showed a similar pattern. The total number of trees/ha decreased sharply during the first fifty years and reached a stable value by the end of the first hundred years. Successional patterns and species competitive abilities were interpreted from the accumulated biomass of each species over the simulation period under the different treatments. Species composition of the stand under the normal treatment showed dominance of deciduous species at early stages of succession, with conifer species being increased and becoming dominant at the end of the simulation period. When the climate was raised 30% (warmer) over the average, the deciduous species continued their dominance over the course of the simulation period. Other simulated species dynamics also appeared to follow what is known about their successional behaviour in the field.