Alterations in climate (or even the natural variation within the current climate) can affect forest communities by altering the internal processes or by altering the proportions of different species in the forests. Experience in assessing the consequences of major climate change is based mostly on paleo-reconstructions for northern hemisphere forests responding to the climate warming that followed the last ice age. These reconstructions demonstrate that climate change also can alter forests by tearing them apart and re-assembling them in novel combinations of species. This process is dramatic in temperate zones and less well-documented but no less certain in the tropics. While the evidence from the past is clear on these points, it is not abundant worldwide. Forming a complete picture of the past is elusive in many tropical regions, even those as prominent as the Amazon, and future climate change may lack past analogs. Our ability to understand the future based on our understanding of what has happened so far in tropical forests therefore faces serious limitations.

Computer modeling of forests can bridge some parts of this gap in understanding. It can be used to explore sets of future climatic conditions that do not currently exist or which have never existed in the history of the Earth. Currently, a wide range of models have been applied to predicting changes in vegetation in response to climate. These models have different data demands and are likely to have rather different domains of applicability. Some have been tested under novel conditions; others draw their credibility by their synthesis of the current best knowledge of ecosystem processes. The models are all flavored by the intended applications and the interests of their developers.

The most frequently applied modeling techniques have their theoretical underpinnings in ecology and ecophysiology. Ecological niche theory provides a framework for understanding two important types of models: species distribution models and gap models. Ecophysiology is relevant to understanding dynamic vegetation models, including the recently developed Earth System Models.

Through an evolving ecological literature, two concepts of niche have emerged. The first involves the factors that control the geographic distributions of species. This concept originated with Joseph Grinnell (1917) and finds substantial application among management-oriented ecologists today (see Shugart, 1998 for a review). It refers to the environmental requirements of species—for example, the range of temperatures a species can withstand—and may be termed environmental niche (Guisan and Thuiller, 2005). An alternate concept of the niche was introduced by Charles Elton (1927) and defined species niches based on feeding relations. This trophic niche concept refers to the way a species obtains and uses resources, especially with respect to other species—for example, a species role within a food web. Over the years, the initial "who eats whom" trophic niche concept was developed to emphasize competitive (as opposed to the predator/prey relations implied by Elton's trophic definition). Hutchinson (1957) attempted a synthesis of the Eltonian and Grinellian niche concepts relating the overlap in environmental requirements (Grinnellian niche) with the likelihood of strong competitive interactions (a post-Eltonian niche concept). In retrospect, this latter development was very important in motivating theory but somewhat less successful at reconciling two rather different concepts of the niche.

Environmental niche theory provides a framework for models that aim to describe species distributions with respect to current and future climate. These models are therefore sometimes referred to as "niche models'' (e.g., Peterson et al., 2002) but are also widely known as "climate envelope'', "bioclimatic", or "species distribution" models (Guisan and Zimmermann, 2000; Guisan and Thuiller, 2005). In these models, a direct or statistical relationship is established between a species known distribution and current climate, and rules to describe this relationship are developed and applied to future climates. Theory would suggest that each species has a total range of environmental tolerance, or fundamental niche, that is greater than the range it actually occupies, or realized niche, due to competition, dispersal history, and other factors—but this distinction is little formalized in models of environmental niche.

Trophic niche theory suggests that species compete for resources and evolve specific resource competition strategies relative to other species. An analogy is sometimes given that the trophic niche is a species' "job" while the environmental niche is its "address". Forest gap models have been developed to simulate resource competition at a small site approximately the size of a tree-fall gap in the forest, using quantitative information about species growth rates and other parameters. Qualitative models including trophic niche interactions have also been developed (e.g., the FATE model; Moore and Noble, 1990). If resource competition determines species survival at a site, it then also influences its distribution at a landscape scale, and these models have begun to be applied on broader scales as computing power has improved.

Ecophysiology theory allows construction of models that describe how plants fix and partition carbon, including the cycling of carbon in soil pools. For some well-studied crop plants, it allows building models of how individual species will respond to specific growth conditions. Tropical forest species are not sufficiently well-studied to be amenable to species-based ecophysiological modeling. However, broad models of carbon partitioning are built from first principles and can be applied in all regions of the world. These models provide information on plant growth forms (or "plant-functional type") that may dominate at a particular site under specified climatic conditions and disturbance regimes. These models are known as "dynamic global vegetation models" (DGVMs). They can be important in simulating distributions of tropical biomes—such as savanna, dry forest, and wet forest. Simplified versions of these models may be integrated into models of global climate. The resulting Earth System Models provide important insight into the interactions of global vegetation changes on climate. Changes in tropical forests have proven to be especially important in this regard.

There are at least two outstanding challenges in modeling tropical forest ecosystems under changing conditions. One challenge involves best including what we know about the photosynthesis process at a cellular level or leaf level in a model. Models constructed to be tested against micrometeorological "flux tower" measurements are likely to include fairly detailed scaled-up leaf processes and attempt to simulate the carbon dioxide, heat, and water fluxes of a forest canopy at a fine temporal resolution. Typically, models that represent the interactions of the planet's surface with global models of climate (DGVMs and Earth System Models) are strongly oriented toward this first challenge.

The second challenge involves predicting the change in structure and composition of vegetation in response to climate changes. In models oriented toward this challenge, the life history attributes of species expressed as parameters for birth, death, and success of species is a central focus. There is also an explicit recognition of the differences in structure of the forest as an overall feature influencing the long-term dynamics of vegetation. Because these models seek to represent the structure of the forest, they better estimate longer-term processes including decomposition and death aspects of forests than models emphasizing the first challenge.

At this time, the problem of simultaneously meeting both of the challenges to model under altered conditions is far from solved—at least with a universally accepted solution (Shugart, 1998). This may stem from the rather large differences between the fine temporal and spatial scales over which photosynthesis is studied compared with the much coarser scale of understanding compositional and structural forest dynamics (Woodward, 1987). A prudent way to proceed is to better understand how different models of forest systems are composed, tested, and applied. Such knowledge is key to interpreting the results from any of the different modeling approaches. Promotion of this understanding is the objective of the review that we present here.

0 0

Post a comment