Versatile process-oriented ecosystem models are discussed as promising tools for the analyses of ecosystem services beyond wood yield, such as catchment water yield, sequestration of carbon and greenhouse gas balances. However, long-term yield simulation is often regarded as a weakness of such versatile models. In this context, we present a multiple response evaluation of the modular, process-based forest growth model MoBiLE-PDT based on mensurational data from 38 permanent sample plots in commercial Eucalyptus globulus plantations in Australia followed from establishment to 8 years of stand age. MoBiLE-PDT is based on the PnET-N-DNDC model and considers nitrogen availability and drought stress dynamically in dependence on tree and stand properties as well as on climate and deposition. New tree dimensions are calculated directly from carbon allocated to sapwood and mortality is derived from stand density. Towards the end of the rotation, model efficiency E was 0.58 for stand volume (m3 ha−1) and 0.54 for aboveground biomass (t C ha−1). In a comparison with similar forest growth models evaluated against the same data only one had a better model efficiency, whereas MoBiLE-PDT was the most versatile model for the analyses of ecosystem services. Due to its modular structure, further model extensions for more ecological applications are easily possible.

In forest management and ecological research, consideration of the impacts and risks of climate change or management optimisation is complex. Computer models have long been applied as tools for these tasks. Process-based forest growth models claim to overcome the limitations of empirical statistical models, but the capacity of different process-based models and modelling approaches have rarely been compared directly. This study evaluates stepwise multiple regression models in comparison to four process-based modelling approaches (3-PG, 3-PG+, CABALA and Forest-DNDC) for greenfield predictions of Eucalyptus globulus plantation growth from 2 to 8 years after planting throughout southern Australia.The stepwise multiple regression models could not simulate plantation growth to a satisfactory level of precision over the entire simulation period, although 2 years after planting model efficiency was 0.46, greater than for any of the process-based models. The variables that were statistically important to predict early plantation growth were mean minimum temperature, stocking rate and the amount of applied N fertiliser. For plantations between 4 and 8 years of age, coefficients of model efficiency were between -0.55 and -0.99.Only process-based models provided the flexibility to realistically predict the impacts of the different growth conditions throughout of the simulation period. Amongst the process-based models, Forest-DNDC achieved the greatest model efficiency (0.28) at 2 years after planting and was the most consistent performing model (0.20-0.30), whereas CABALA achieved the greates model efficiency (0.70) after 8 years of plantation growth. Both 3-PG models performed best for 6-year-old plantations, but throughout the simulation period their predictive precision was strongly dependent on estimating site fertility. For 3-PG+, a statistical approach to estimate site fertility resulted in a model efficiency of 0.28 at 6 years after planting, whereas a subjective estimation of site fertility for 3-PG resulted in a model efficiency of 0.58. In general, the process-based models had difficulties in simulating very young plantations at less then 4 years after planting, plantations with high tree mortality rates and plantation response to extreme silvicultural management operations. (C) 2008 Elsevier B.V. All rights reserved.

Soils provide the largest terrestrial carbon store, the largest atmospheric CO2 source, the largest terrestrial N2O source and the largest terrestrial CH4 sink, as mediated through root and soil microbial processes. A change in land use or management can alter these soil processes such that net greenhouse gas exchange may increase or decrease. We measured soil–atmosphere exchange of CO2, N2O and CH4 in four adjacent land‐use systems (native eucalypt woodland, clover‐grass pasture, Pinus radiata and Eucalyptus globulus plantation) for short, but continuous, periods between October 2005 and June 2006 using an automated trace gas measurement system near Albany in southwest Western Australia. Mean N2O emission in the pasture was 26.6 μg N m−2 h−1, significantly greater than in the natural and managed forests (< 2.0 μg N m−2 h−1). N2O emission from pasture soil increased after rainfall events (up to 100 μg N m−2 h−1) and as soil water content increased into winter, whereas no soil water response was detected in the forest systems. Gross nitrification through 15N isotope dilution in all land‐use systems was small at water holding capacity < 30%, and under optimum soil water conditions gross nitrification ranged between < 0.1 and 1.0 mg N kg−1 h−1, being least in the native woodland/eucalypt plantation < pine plantation < pasture. Forest soils were a constant CH4 sink, up to −20 μg C m−2 h−1 in the native woodland. Pasture soil was an occasional CH4 source, but weak CH4 sink overall (−3 μg C m−2 h−1). There were no strong correlations (R < 0.4) between CH4 flux and soil moisture or temperature. Soil CO2 emissions (35–55 mg C m−2 h−1) correlated with soil water content (R < 0.5) in all but the E. globulus plantation. Soil N2O emissions from improved pastures can be considerable and comparable with intensively managed, irrigated and fertilised dairy pastures. In all land uses, soil N2O emissions exceeded soil CH4 uptake on a carbon dioxide equivalent basis. Overall, afforestation of improved pastures (i) decreases soil N2O emissions and (ii) increases soil CH4 uptake.

Linking environmental computer simulation models and geographic information systems (GIS) is now a common practice to scale up simulations of complex ecosystem processes for decision support. Unfortunately, several important issues of upscaling using GIS are rarely considered; in particular scale dependency of models, availability of input data, support of input and validation data, and uncertainty in prediction including error propagation from the GIS. We linked the biogeochemical Forest-DNDC model to a GIS database to predict growth of Eucalyptus globulus plantations at two different scales (similar to 0.045 ha plot(-1) scale and similar to 100 ha grid(-1) scale) across Victoria, in south-eastern Australia. Results showed that Forest-DNDC was not scale dependent across the range of scales investigated. Reduced availability of input data at the larger scale may introduce severe prediction errors, but did not require adjustment of the model in this study. Differences in the support of input and validation data led to an underestimation of predictive precision but an overestimation of prediction accuracy. Increasing data support, produced a high level of prediction accuracy ((-)e%), but a medium level of predictive precision (r(2)=0.474, ME=0.318) after statistical validation. GIS error contribution could be detected but was not readily or reliably quantified. In a regional case study for 2653 ha of E. globulus plantations, the linked model GIS system estimated a total standing biomass of 95 260 t C for mid-2003 and a net CO2 balance of -45 671 t CO2-C yr(-1) for the entire year of 2002. This study showed that regional predictions of forest growth and carbon sequestration can be produced with greater confidence after a comprehensive assessment of upscaling issues.

The tree growth sub-module (PnET) of the mechanistic model Forest-DNDC was calibrated and validated for plantation grown Eucalyptus globulus. Forest-DNDC describes the biogeochemical cycles of C and N and can assist in estimating soil-borne greenhouse gas fluxes. For validation of the forest growth sub-module, data from commercial forest plantations in south-eastern Australia was used. Growth predictions agreed well with growth measurements taken at age 6 years from 28 permanent sample plots, with an average prediction error of −1.62tCha−1 (−3.19%). Differences between predicted and measured aboveground C stocks ranged between −23.5 and 12.6tCha−1, which amounted to a relative root mean square error in prediction of 17.9%. Correlation between modelled and measured C in standing biomass was good (r2=0.73), with a Nash–Sutcliffe coefficient of model efficiency, ME=0.65. The results obtained from the validation test reveal that Forest-DNDC can predict growth of E. globulus to a high level of precision across a broad range of climatic conditions and soil types. Forest-DNDC performed satisfactorily in comparison to other growth and yield models that have already been calibrated for E. globulus (e.g. BIOMASS, 3-PG, PROMOD or CABALA). In contrast to these growth and yield models, Forest-DNDC can additionally estimate total greenhouse gas budgets. The slightly lower precision of Forest-DNDC in comparison with specific management models, such as CABALA, are compensated for by the simple input requirements and application to regional situations.

The process-based models DNDC and PnET-N-DNDC were evaluated with regard to their potential to calculate regional inventories of N-trace gas emissions from agricultural and forest soils. To extend the model predictions to regional scale, we linked the models to a detailed GIS-database for Saxony, Germany, which was holding all the spatially and temporally differentiated input information and other model drivers. Total annual N2O-emissions from agricultural soils in Saxony ranged from 0.5–26.0 kg N2O-N ha−1 yr−1 and were calculated to amount to approx. 5475 t N2O-N yr−1 in the year 1995, which compares quite well with previous estimates based on the IPCC approach (4892 t N2O-N yr−1). Compared to the agricultural soils, N2O-emissions from forest soils in Saxony (range: 0.04–19.7 kg N2O-N ha−1 yr−1) were much lower and amounted to 1011 t N2O-N yr−1. In comparison with other sources of N2O in Saxony our estimates show, that – even in such a highly industrialised region like Saxony – soils contribute more than 50% to the total regional N2O source strength. Simulated emissions of NO from the agricultural and forest soils were approx. in the same magnitude than for N2O. The modelled NO-emission rates ranged from 0.4–26.3 kg NO-N ha−1 yr−1 for the agricultural soils and 0.04–28.3 kg NO-N ha−1 yr−1 for the forest soils with total emissions of 8868 t NO-N yr−1 (agricultural soils) and 4155 t NO-N yr−1 (forest soils). Our results indicated that the agricultural and forest soils were a significant source, which contributed 17.9% of the total NOx emissions from various sources in Saxony. Furthermore, a series of sensitivity tests were carried out, which demonstrated that variations in soil organic carbon content (SOC) and soil texture significantly effect the modelled N-trace gas emissions from agricultural soils at the regional scale, whereas, in addition, for forest soils also the soil pH is within the sensitive factors. Finally, multi-year simulations were conducted for the region with observed meteorological data from 1994–1996. The results demonstrated that the modelled interannual variations, which were obviously induced by only the climate conditions, in the N-gas emissions were as high as 36%. The high interannual variations imply that multi-year (e.g., 5–10 years), instead of single baseline year, simulations would produce more reliable estimates of mean soil N2O-emissions at regional scale. With respect to the Kyoto protocol this means that the mean N2O-emissions from soils in the period 1988–1992 should be evaluated instead of focusing on a single year, 1990.

It is well known that soils are major sources for the atmospheric N-trace gases NO and N2O. The key processes responsible for N-trace gas production and consumption in soils are the microbial processes nitrification and denitrification. Field measurements have revealed that N-trace gas emissions are both highly variable in time and space. Furthermore, the magnitude of N-trace gas emissions is dependent on land use (e.g. agricultural and forest soils) and land management (e.g. fertilization rate). Due to the complexity of factors affecting the magnitude of soil-born N-trace gas emissions estimates on the contribution of entire landscapes to the regional and global budgets of N2O and NO are still highly uncertain. However, recent developments in process oriented modeling have shown that these models are now capable to catch the patterns of spatial and temporal variations in N-trace gas fluxes for a wide variety of ecosystems and patterns of land use management. This offers the chance to couple such models to GIS-databases -with all relevant information necessary to initialize and drive the models- in order to calculate regional inventories of N-trace gas emissions from complex landscapes, with a hitherto unknown precision. Furthermore, the effect of land-use changes and different strategies of land use management on N-trace gas emissions can be evaluated. Using Saxony, Germany, as a case study area for complex terrain - offering a mixture of agricultural and forest areas in hilly and lowland regions - this paper will demonstrate the strategy, the strengths and weaknesses of the process oriented models DNDC (agricultural soils) and PnET-N-DNDC (forest soils) for calculating regional N-trace gas inventories.