**Maintaining Ecosystem Function by Restoring Forest Biodiversity – Reviewing Decision-Support Tools that link Biology, Hydrology and Geochemistry**

Yueh-Hsin Lo, Juan A. Blanco, Clive Welham and Mike Wang

Additional information is available at the end of the chapter

http://dx.doi.org/10.5772/59390

## **1. Introduction**

Land use change in forest ecosystems is a worldwide problem. In many cases, however, the change is only temporary, and after a period of economic activity, the original forest must be reclaimed back to its original (or as close as possible) estate. A typical case is in open-pit mining. In many juridictions there is a legal requirement for the company to engage in restorative activities designed to bring back biodiversity and function to those areas espoiled by mining.

To reclaim a fully functional forest ecosystem, soil, topographic and hydrological properties must foster the biogeochemical and ecological processes required to support a vigorous vegetative community. While significant advances have been made in this regard, each reclaimed forest ecosystem is unique, and there remains considerable uncertainty as to how these interdependent processes will be manifest in any particular instance. One of the most important interdepencies is between water availability and plant uptake. Our understanding of biodiversity and linkage with surface water availability and distribution is limited because this relationship has not been examined within and across scales. The availability and distri‐ bution of water can influence ecosystem structure and function at a range of scales and levels of organization through its influences on various processes and feedbacks that can affect both animals and plants. For example, at a landscape level, the distribution of herbivore home ranges and vegetation communities may be influenced by the mosaic created by water sources. At the ecosystem and community levels, ecosystem processes such as nutrient cycling, predator-prey interactions, and interspecific competition may be influenced by the availablil‐ ity and location of water sources. At a population level, surface and soil water availability and

© 2015 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. © 2015 The Author(s). Licensee InTech. This chapter is distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/3.0), which permits unrestricted use, distribution, and eproduction in any medium, provided the original work is properly cited.

distribution may influence herbivore and vegetation survivorship through processes such as droughts, pests and diseases.

Water scarcity can be produced by seasonal or annual droughts, but also by difficulties in uptake due to salinity or other contaminants. Naturally saline systems within the boreal forest are infrequent but are widespread across western North America [1]. Typical saline sites can be found in partially-closed catchment areas with inflows of groundwater [2]. Levels of salinity in shallow soils generally increase as the elevation declines toward a basin, reflecting the movement of the salts along the topographic gradient.

Reclaimed landscapes must hold enough water for plant growth but have enough downward movement of water to flush any contaminants from the rooting zone. Excessive percolation is also a concern, however, since drainage from toe slopes can carry dissolved contaminates out of a landform [3]. In northern climates, evapotranspiration is the largest annual loss of water [4], and a primary factor in the movement of salts to the surface. Interactions between vege‐ tation productivity, nutrient budgets and evapotranspiration must therefore be assessed simultaneously [5]. There has been little research, however, on how changes in vegetation and climate affect the energy balance and water movement in northern ecosystems [3].

A broad array of decision support tools is available for projecting forest stand development in reclaimed ecosystems. These include forest ecosystem classification and ecosite manuals, stand establishment keys/guides, competition index methodologies, volume tables, site index curves, soil-site equations, stand density management tools, and growth and yield equations/ tables/decision systems [6]. Many of these tools are empirically based. This means they have significant limitations in their ability to accurately project productivity because in a strict sense, their application should be restricted to the stand conditions from which their underlying relationships were derived [7]. This can be problematic in mine reclamation for two reasons. First, the soil prescriptions that form the basis for reclamation often lack the historical legacy (propagule bank, organic matter, soil structural and biochemical properties, etc.) common to natural sites [8-11]. Secondly, forests are growing under climatic conditions that differ from the historical climate regime, particularly in more northerly regions. Global circulation model projections indicate continued increases in atmospheric and surface temperatures at least through this century, along with associated changes in the precipitation regime. Historical properties are now no longer tenable as the sole basis for deriving empirical growth relation‐ ships.

With the widespread increase in computing power, model sophistication and complexity have seen a rapid increase though this has not necessarily resulted in better and more reliable outcomes [12]. One issue with greater complexity is the increased cost and difficulty of obtaining calibration data sets for specific local ecosystems. From the perspective of reclama‐ tion, another issue is that most forest stand decision support tools have little or no represen‐ tation of hydrological processes. In essence, the implicit assumption underlying these models is that the hydrological regime is in an equilibrium condition such that short-term temporal or spatial fluctuations in available moisture are of little consequence to long-term trends in productivity.

One way of resolving these limitations is to build a fully integrated vegetation-hydrology tool that combines a detailed representation of all critical processes within a single computing environment. The resulting 'mega-model' would possess the benefits of a fully integrated and interactive system with appropriate feedbacks and system controls. In such models, linkage between the different components of the model would be in real time, with vegetation growth and development in each time step related to the hydrological processes driving available moisture [13].

distribution may influence herbivore and vegetation survivorship through processes such as

Water scarcity can be produced by seasonal or annual droughts, but also by difficulties in uptake due to salinity or other contaminants. Naturally saline systems within the boreal forest are infrequent but are widespread across western North America [1]. Typical saline sites can be found in partially-closed catchment areas with inflows of groundwater [2]. Levels of salinity in shallow soils generally increase as the elevation declines toward a basin, reflecting the

Reclaimed landscapes must hold enough water for plant growth but have enough downward movement of water to flush any contaminants from the rooting zone. Excessive percolation is also a concern, however, since drainage from toe slopes can carry dissolved contaminates out of a landform [3]. In northern climates, evapotranspiration is the largest annual loss of water [4], and a primary factor in the movement of salts to the surface. Interactions between vege‐ tation productivity, nutrient budgets and evapotranspiration must therefore be assessed simultaneously [5]. There has been little research, however, on how changes in vegetation and

A broad array of decision support tools is available for projecting forest stand development in reclaimed ecosystems. These include forest ecosystem classification and ecosite manuals, stand establishment keys/guides, competition index methodologies, volume tables, site index curves, soil-site equations, stand density management tools, and growth and yield equations/ tables/decision systems [6]. Many of these tools are empirically based. This means they have significant limitations in their ability to accurately project productivity because in a strict sense, their application should be restricted to the stand conditions from which their underlying relationships were derived [7]. This can be problematic in mine reclamation for two reasons. First, the soil prescriptions that form the basis for reclamation often lack the historical legacy (propagule bank, organic matter, soil structural and biochemical properties, etc.) common to natural sites [8-11]. Secondly, forests are growing under climatic conditions that differ from the historical climate regime, particularly in more northerly regions. Global circulation model projections indicate continued increases in atmospheric and surface temperatures at least through this century, along with associated changes in the precipitation regime. Historical properties are now no longer tenable as the sole basis for deriving empirical growth relation‐

With the widespread increase in computing power, model sophistication and complexity have seen a rapid increase though this has not necessarily resulted in better and more reliable outcomes [12]. One issue with greater complexity is the increased cost and difficulty of obtaining calibration data sets for specific local ecosystems. From the perspective of reclama‐ tion, another issue is that most forest stand decision support tools have little or no represen‐ tation of hydrological processes. In essence, the implicit assumption underlying these models is that the hydrological regime is in an equilibrium condition such that short-term temporal or spatial fluctuations in available moisture are of little consequence to long-term trends in

climate affect the energy balance and water movement in northern ecosystems [3].

droughts, pests and diseases.

1442 Biodiversity in Ecosystems - Linking Structure and Function

ships.

productivity.

movement of the salts along the topographic gradient.

Seldom it is feasible or desirable, however, to build a model that includes all processes and scales of interest [14]. There are no a single model can be used in every situation that could arise during planning or management. If such a model is constructed, it will probably have one or more issues of being too generalist, having issues to represent processes at different scales, and continuous tuning and modifications to handle new data and issues [15]. Further‐ more, ensuring these models are reliable requires substantial investments of time and energy, and their applicability tends to become overly specialized thereby reducing flexibility and portability (applying the model in different locations or circumstances). Other disadvantages include an increased calibration load associated with the myriad feedback and system controls, validation (establishing the veracity of model structure and architecture) and verification (assessing output accuracy). In this respect, more complex models are not 'better' due simply to the fact they incorporate a greater representation of reality because complexity does not necessarily equate to improved accuracy and precision. This leads to the modelling mantra that models should be only as complex as absolutely necessary.

A compromise (and popular) approach is to construct a 'meta-model' wherein the most suitable forest productivity and hydrological models would be coupled as input-output (I/O) systems. This I/O linkage refers to the idea that output from a given model serves as input to another. Fall [14] provided a list of the benefits and costs of the meta-modeling approach. The most important benefits of the meta-modelling approach are that it can use previous knowl‐ edge and expertise generated when developed well document models, but at the same time allow for flexibility to match the meta-model to the local conditions, data availability and other user needs. In addition, different teams can work in different processes and sub-models at the same time, improving the use of time and resources. Such advantage is particularly important when individuals are separated geographically. Another advantage is that by linking different models or sub-models, each of them can be analyzed and validated separately. Such advantage is important to increase understanding of complex interactions between different ecosystem components, and to allow comparisons of different model components. Data flow in the metamodelling approach is also more flexible, and output from one model can be used as input for several other model components. This allows partial verification as intermediate data can be analyzed and stored, something important in adaptive management and monitoring. Finally, in a meta-modelling framework model sensitivity and scenario analyses are facilitated as they can be performed for different model components.

Not all forest models are applicable to a meta-modelling approach. Hence, the objective of the research presented here was to identify and compare the available forest models already being used in research, and to evaluate their suitability for use as decision-support tools in designing successful restoration plans to bring forest biodiversity and function back to sites disturbed by industrial activities (mining in particular).

## **2. Review methodology**

### **2.1. Literature review on forest growth models**

The review covered papers available at the beginning of 2013 in scientific journals, scientific books, proceedings of scientific workshops and conferences, and technical reports. The literature search was conducted using the term "forest growth model" in combination with each of the following 8 keywords: "climate change, gap model, hydrology, mixedwood, productivity, regeneration, simulation, and succession". 'Hits' were then screened and those pertaining to ecosystems other than temperate and boreal forests were eliminated (tropical and subtropical ecosystems, Mediterranean ecosystems, grasslands, etc.). The data bases consulted were:

**•** *Canadian Forest Service Bookstore* (list of publication by Canadian authors compiled by the Canadian Forest Service compiled by the Canadian Forest Service)

Available at http://bookstore.cfs.nrcan.gc.ca

**•** *ISI Web of knowledge* (academic search engine by Thomson Reuters).

Available at http://www.isiwebofknowledge.com

**•** *Google Scholar* (academic search engine by Google Inc.).

Available at http://scholar.google.ca

**•** *C.E.M.A. Research library* (collection of reports and publication for the Alberta Oil Sands area compiled by the Cumulative Environmental Management Association)

Available at http://www.cemaonline.ca


Available at http://ecobas.org/www-server/index.html

### **2.2. Ranking of forest models**

Development of a system for ranking models is a challenging and inexact exercise. The criterion used to build the ranking system, for example, as well as the relative weighting attached to each ranking variable are important decisions. First and foremost, we are of the opinion that the scientific peer review process provides the best assurance that model structure and application are sound and have been subjected to expert scrutiny, particularly for those models with multiple entries in the scientific literature. Hence, only models cited in three or more peerreviewed publications were considered in creating a ranking of model suitability. Models with lesser publications were therefore omitted from further analysis because their low publication rates indicate that they have not yet received a proper assessment of their suitability. The objective of the ranking exercise was to identify the best 3 to 5 models according to five criteria. The ranking was created in two steps:

### *2.2.1. Initial partial score (0 to 8 points)*

successful restoration plans to bring forest biodiversity and function back to sites disturbed

The review covered papers available at the beginning of 2013 in scientific journals, scientific books, proceedings of scientific workshops and conferences, and technical reports. The literature search was conducted using the term "forest growth model" in combination with each of the following 8 keywords: "climate change, gap model, hydrology, mixedwood, productivity, regeneration, simulation, and succession". 'Hits' were then screened and those pertaining to ecosystems other than temperate and boreal forests were eliminated (tropical and subtropical ecosystems, Mediterranean ecosystems, grasslands, etc.). The data bases

**•** *Canadian Forest Service Bookstore* (list of publication by Canadian authors compiled by the

**•** *C.E.M.A. Research library* (collection of reports and publication for the Alberta Oil Sands area

**•** *Register Of Ecological Models* (self-registration tool to compile ecological models by the Kasel

Development of a system for ranking models is a challenging and inexact exercise. The criterion used to build the ranking system, for example, as well as the relative weighting attached to each ranking variable are important decisions. First and foremost, we are of the opinion that the scientific peer review process provides the best assurance that model structure and

Canadian Forest Service compiled by the Canadian Forest Service)

**•** *ISI Web of knowledge* (academic search engine by Thomson Reuters).

compiled by the Cumulative Environmental Management Association)

**•** *U.B.C. library* (academic library at the University of British Columbia)

by industrial activities (mining in particular).

1464 Biodiversity in Ecosystems - Linking Structure and Function

**2.1. Literature review on forest growth models**

Available at http://bookstore.cfs.nrcan.gc.ca

Available at http://scholar.google.ca

Available at http://www.cemaonline.ca

Available at http://www.isiwebofknowledge.com

**•** *Google Scholar* (academic search engine by Google Inc.).

**•** On-line catalogue available at http://www.library.ubc.ca

Available at http://ecobas.org/www-server/index.html

**2. Review methodology**

consulted were:

University).

**2.2. Ranking of forest models**

Each model was scored initially using four criteria that varied in maximum value between 1.5 to 2.5 points. A proportional approach was used to derive a relative ranking for each model within a given criterion: The model with the best mark received the maximum criterion score, and the remainder were scored in direct proportion to how they compared with the top model. The four criteria used to create the rankings were:


A final partial score for a given model was calculated by adding up the score for each criterion.

### *2.2.2. Full score (0 to 10 points) and shortlisting*

An additional criterion was defined as the number of countries, ecosystem types and forest types in which each model had been applied. This criterion was considered a measure of model versatility, and was assigned a potential maximum of 2 points. Given the time-consuming nature of gathering data to calculate the values of this criterion, only models with an accumu‐ lated score of 5 or above (out of a maximum score of 8) for the previous four criteria were given scores for model versatility. Following the ranking exercise, a total score (out of a maximum of 10 points) was calculated and those models with 6.5 points or more made the shortlist.

All models can be broadly classified into three categories, depending on their structure and how they are parameterized:


**•** *Hybrid models:* combine elements of the above two categories. In this approach, empirical data are used to parameterize one or more of the ecophysiological processes driving in tree growth and ecosystem production.

## **3. Results and discussion**

### **3.1. Literature review**

A total of 466 documents were identified through the literature search. Most of the documents were detected using the general modeling-related keywords: "gap model" (83 documents), "productivity" (89 documents), "regeneration" (86 documents), and "simulation" (33 docu‐ ments). These 291 documents account for 62.3% of the total. Following the pioneering work of Botkin et al. [16] with development of the JABOWA gap model, the number of models addressing forest productivity has increased steadily over the previous four decades (Figure 1). Several factors have likely contributed to this trend. Most models developed prior to the 1990s were designed to simulate timber production. Since then, forest management has moved from an almost exclusive focus on timber towards an emphasis on the sustainable production of multiple ecosystem goods and services [7]. Subsequent model developments have reflected this change.

**Figure 1.** Number of documents published in a given year as derived from the keyword searches.

Climate and climate change have also emerged as a dominant issue in forest management, as government and industry strive to understand its impact on the present and future flow of goods and services. Keywords related to moisture ("climate change", "hydrology") accounted for 96 documents, 20.8% of the total. Of the 96 documents, 33 were related to the "climate change' keyword and 63 under "hydrology". A proportion of the documents under the keyword "hydrology" were also related to how climate change might alter the hydrological cycle. To avoid duplication, those documents were not accounted under the results for "climate change". Hence, under the "climate change" keyword only documents that deal with climate change and anything other than hydrological process are accounted for (mostly temperaturerelated research).

**•** *Hybrid models:* combine elements of the above two categories. In this approach, empirical data are used to parameterize one or more of the ecophysiological processes driving in tree

A total of 466 documents were identified through the literature search. Most of the documents were detected using the general modeling-related keywords: "gap model" (83 documents), "productivity" (89 documents), "regeneration" (86 documents), and "simulation" (33 docu‐ ments). These 291 documents account for 62.3% of the total. Following the pioneering work of Botkin et al. [16] with development of the JABOWA gap model, the number of models addressing forest productivity has increased steadily over the previous four decades (Figure 1). Several factors have likely contributed to this trend. Most models developed prior to the 1990s were designed to simulate timber production. Since then, forest management has moved from an almost exclusive focus on timber towards an emphasis on the sustainable production of multiple ecosystem goods and services [7]. Subsequent model developments have reflected

**Figure 1.** Number of documents published in a given year as derived from the keyword searches.

Climate and climate change have also emerged as a dominant issue in forest management, as government and industry strive to understand its impact on the present and future flow of goods and services. Keywords related to moisture ("climate change", "hydrology") accounted for 96 documents, 20.8% of the total. Of the 96 documents, 33 were related to the "climate change' keyword and 63 under "hydrology". A proportion of the documents under the

growth and ecosystem production.

1486 Biodiversity in Ecosystems - Linking Structure and Function

**3. Results and discussion**

**3.1. Literature review**

this change.

Tian et al. [17] argue that current forest growth models are seldom "well balanced" in terms of equivalence in the detail with which water, C, and N cycles are represented (see also [18, 19]). For instance, G'DAY [20], PnET-CN [21], and Biome-BGC [22] simulate forest growth and/ or biogeochemical processes in detail but are much less rigorous in their approach to repre‐ senting forest hydrology. Unbalanced model design is likely to limit the ability of a given model to accurately predict the hydrology and biogeochemistry of forest ecosystems in response to changes in climate, land use, and/or management practices [19, 23].

In recent years, there have been several attempts to better link forest growth with hydrology. Chen and Driscoll [24] demonstrated that incorporating a more detailed hydrologic cycle into the Biome-BGC model improved predictions of seasonal effluent nitrate concentrations. Seely et al. [25] developed a stand-alone hydrology model for forest management applications (ForWaDy), with the explicit objective of minimizing data requirements. This model has been incorporated into the forest ecosystem simulation model, FORECAST [7]. Evidence suggests it provides a robust representation of moisture availability on tree growth, based on the balance between inputs from precipitation and seepage, and outputs by canopy interception, evapo‐ transpiration, plant uptake, percolation and runoff [26].

**Figure 2.** Number of document per model, for models with four or more documents in the database.

Despite the large number of models identified from the initial search, only 22 had 3 or more references (Figure 2). This suggests that many models are developed as one-time tools to explore issues of scientific importance, rather than as decision-support tool in support of management. This can limit the ease with which a model can be applied to situations different from that for which it was originally designed (i.e., the model´s portability). It might also constrain the flexibility in model architecture, making it difficult to add management capa‐ bilities at a later date [7].

### **3.2. Model shortlisting**

Only five models had more than 10 references: ECOSYS (18), FORECAST (17), 3-PG (15), BGC (13), and SORTIE (11; including both of its versions, SORTIE-ND, SORTIE-BC). It should be noted that the count for BGC is inflated by the fact it has three different variants (TREE-BGC, FOREST-BGC, and BIOME-BGC) which were grouped together for purposes of analysis. Arguably, it may be more appropriate to consider each separately since they are applicable to widely different scales (tree, stand, and biome, respectively). In that case, the ranking for each separate model would be much lower. The models LINKAGES, BIOMASS and CENTURY had relatively few publications but they ranked fairly well in the remaining criteria (Table 1). In the case of ECOSYS, 14 of the 18 references listed the developer (Grant) as the primary author. This is an unusually high number despite the fact the model was first published 14 years ago (Table 1). Publications of the remaining top models in the review encompassed a broad range of authors. This could be an indication that ECOSYS has not received broad acceptance, which could at least partly explain its low citation rate (see below). The common features of the top seven models (in green and yellow colours in Table 1) are that they are subject to ongoing development (≥ 14 years), have been broadly applied in temperate and boreal ecosystems, and been cited within the last 2 years. The one exception is BIOMASS, which has not been cited in the previous 4 years. The top models were also heavily cited, with more 80 citations; ECOSYS was the clear exception, with only 25 citations (Table 1).

On its own, a high citation rate alone does not necessarily mean that a model is indeed being used extensively or is pertinent to the needs of Total. For example, a model that has been cited extensively is CENTURY. This model was originally developed for grasslands and so refer‐ ences to CENTURY were relatively frequent in the agricultural literature. In its current version, the model possesses a crude ability to represent forest growth. Its focus, however, is still mainly on soil processes even though this may be within the context of forest management. To an extent our ranking system was designed to take these factors into account by assigning the publication rate a higher weighting than the citation rate (a maximum score of 2.5 versus 2, respectively).

An important aspect of model suitability to oil sands reclamation is its portability. Portability refers to the ease with which a model can be calibrated, and its algorithms applied to, an ecosystem different from that in which it was originally developed. This is because no tool has been developed specifically for the conditions that characterize oil sands materials. Hence, the higher the number of countries and ecosystems where the model has been successfully applied, the higher its portability. The portability criterion was the final key factor that discriminated among the "shortlisted" models (BGC, FORECAST, 3-PG, and ECOSYS) and the remainder (Figure 2). These four models had more than 10 documents in the database, meaning they have been used in more than 10 different countries and/or ecosystem types, indicating a high portability potential.

### *3.2.1. ECOSYS*

explore issues of scientific importance, rather than as decision-support tool in support of management. This can limit the ease with which a model can be applied to situations different from that for which it was originally designed (i.e., the model´s portability). It might also constrain the flexibility in model architecture, making it difficult to add management capa‐

Only five models had more than 10 references: ECOSYS (18), FORECAST (17), 3-PG (15), BGC (13), and SORTIE (11; including both of its versions, SORTIE-ND, SORTIE-BC). It should be noted that the count for BGC is inflated by the fact it has three different variants (TREE-BGC, FOREST-BGC, and BIOME-BGC) which were grouped together for purposes of analysis. Arguably, it may be more appropriate to consider each separately since they are applicable to widely different scales (tree, stand, and biome, respectively). In that case, the ranking for each separate model would be much lower. The models LINKAGES, BIOMASS and CENTURY had relatively few publications but they ranked fairly well in the remaining criteria (Table 1). In the case of ECOSYS, 14 of the 18 references listed the developer (Grant) as the primary author. This is an unusually high number despite the fact the model was first published 14 years ago (Table 1). Publications of the remaining top models in the review encompassed a broad range of authors. This could be an indication that ECOSYS has not received broad acceptance, which could at least partly explain its low citation rate (see below). The common features of the top seven models (in green and yellow colours in Table 1) are that they are subject to ongoing development (≥ 14 years), have been broadly applied in temperate and boreal ecosystems, and been cited within the last 2 years. The one exception is BIOMASS, which has not been cited in the previous 4 years. The top models were also heavily cited, with more 80 citations; ECOSYS

On its own, a high citation rate alone does not necessarily mean that a model is indeed being used extensively or is pertinent to the needs of Total. For example, a model that has been cited extensively is CENTURY. This model was originally developed for grasslands and so refer‐ ences to CENTURY were relatively frequent in the agricultural literature. In its current version, the model possesses a crude ability to represent forest growth. Its focus, however, is still mainly on soil processes even though this may be within the context of forest management. To an extent our ranking system was designed to take these factors into account by assigning the publication rate a higher weighting than the citation rate (a maximum score of 2.5 versus 2,

An important aspect of model suitability to oil sands reclamation is its portability. Portability refers to the ease with which a model can be calibrated, and its algorithms applied to, an ecosystem different from that in which it was originally developed. This is because no tool has been developed specifically for the conditions that characterize oil sands materials. Hence, the higher the number of countries and ecosystems where the model has been successfully applied, the higher its portability. The portability criterion was the final key factor that discriminated among the "shortlisted" models (BGC, FORECAST, 3-PG, and ECOSYS) and the remainder (Figure 2). These four models had more than 10 documents in the database, meaning they have

bilities at a later date [7].

1508 Biodiversity in Ecosystems - Linking Structure and Function

**3.2. Model shortlisting**

respectively).

was the clear exception, with only 25 citations (Table 1).

ECOSYS is a process-based, ecosystem-level model. It was originally developed as a soil model for agricultural ecosystems, but since then it has evolved into a complex simulator of the plantatmosphere-soil system [27]. ECOSYS has a time step of one hour. It has representation of multi-layered canopies and soils. In this model, flows and transformation of growth resources (radiation, water, C, N, and P) are simulated for populations of plants and microorganisms. The model is constructed in order to link at different spatial and biological scales ecological processes that determine the ecophysiology of linked plant and microbial populations.

The model can represent ecosystems scales from homogeneous stands to heterogeneous landscapes, including natural and human-made disturbances. The model estimates ecosystem productivity through an energy balance approach.

Energy flows are simulated between the atmosphere and ground surfaces (snow, soil, litter). For plants, energy flows are simulated between the atmosphere and leaf or stem surfaces [29]. To calculate total exchange energy, energy exchanges between all plant and ground surfaces are added up. Hydrological processes (surface runoff, infiltration macro- and micro-pore flow) are then coupled with surface energy exchange and soil heat transfer [28].

ECOSYS calculates energy exchange in the canopy at an hourly basis, using a two-stage convergence solution to estimate heat and water transfers for the soil-root-canopy system for several plant populations and layers of soil and canopy. In the first stage, a canopy temperature value is calculated for each plant population by closing the canopy energy balance (sensible heat, latent heat flux, net radiation, and change in stored heat).These fluxes are controlled by aerodynamic and canopy stomatal resistances [29].

The simulation of water status effects on energy exchange is based on coupling the uptake of water from the soil through the root to the canopy, with the evaporation of water from the canopy to the atmosphere. This coupling determines the water status of the canopy and hence its conductance to water vapor [27].

Leaf C fixation is determined by carboxylation, which is controlled by irradiance, temperature, and leaf CO2 concentration, and by diffusion, which is controlled by the atmosphere-leaf CO2 concentration gradient and leaf conductance. The coupling of carboxylation and diffusion in ECOSYS allows the calculation of a leaf C fixation rate, which is then aggregated to the canopy level. Net C exchange between plants and the atmosphere is the difference between the two. Losses of leaf C are accelerated by reducing C fixation compared to maintenance respiration by reduced availability of growth resources (N, heat, or water). Net CO2 fixation is calculated for each branch as the difference between gross fixation and the sum of respiration through maintenance, growth, and reproduction [27].

