Use of Coarse‐Resolution Models of Species’ Distributions to Guide Local Conservation Inferences

Abstract: Distribution models are used increasingly for species conservation assessments over extensive areas, but the spatial resolution of the modeled data and, consequently, of the predictions generated directly from these models are usually too coarse for local conservation applications. Comprehensive distribution data at finer spatial resolution, however, require a level of sampling that is impractical for most species and regions. Models can be downscaled to predict distribution at finer resolutions, but this increases uncertainty because the predictive ability of models is not necessarily consistent beyond their original scale. We analyzed the performance of downscaled, previously published models of environmental favorability (a generalized linear modeling technique) for a restricted endemic insectivore, the Iberian desman (Galemys pyrenaicus), and a more widespread carnivore, the Eurasian otter (Lutra lutra), in the Iberian Peninsula. The models, built from presence–absence data at 10 × 10 km resolution, were extrapolated to a resolution 100 times finer (1 × 1 km). We compared downscaled predictions of environmental quality for the two species with published data on local observations and on important conservation sites proposed by experts. Predictions were significantly related to observed presence or absence of species and to expert selection of sampling sites and important conservation sites. Our results suggest the potential usefulness of downscaled projections of environmental quality as a proxy for expensive and time‐consuming field studies when the field studies are not feasible. This method may be valid for other similar species if coarse‐resolution distribution data are available to define high‐quality areas at a scale that is practical for the application of concrete conservation measures.     

The Effectiveness of Iberian Protected Areas in Conserving Terrestrial Biodiversity

Abstract:  The Iberian Peninsula harbors about 50% of European plant and terrestrial vertebrate species and more than 30% of European endemic species. Despite the global recognition of its importance, the selection of protected areas has been ad hoc and the effectiveness of such choices has rarely been assessed. We compiled the most comprehensive distributional data set of Iberian terrestrial plant and vertebrate species available to date and used it to assess the degree of species representation within existing protected areas. Existing protected areas in Spain and Portugal reasonably represented the plant and animal species we considered (73–98%). Nevertheless, species of some groups (amphibians, reptiles, birds, and gymnosperms) did not accumulate in protected areas at a rate higher than expected by chance ( p > 0.05). We determined that to conserve all vertebrate and plant species in the Iberian Peninsula, at least 36 additional areas are needed. Selection of additional areas for conservation would be facilitated if such areas coincided with sites of community importance (SCI) designated under the European Commission Habitats Directive. Additional areas required for full representation of the selected plant and animal species all coincide with SCI in Spain. Nevertheless, the degree of coincidence varies between 0.3% and 74.6%, and there is a possibility that important areas for conservation occur outside the SCI. Our results support the view that current SCI can be used for prioritization of areas for conservation, but a systematic reevaluation of conservation priorities in Spain and Portugal would be necessary to ensure that effective conservation of one of European's most important biodiversity regions is achieved.     

Application of a Random Forest algorithm to predict spatial distribution of the potential yield of Ruditapes philippinarum in the Venice lagoon, Italy

We present a modelling framework that combines machine learning techniques and Geographic Information Systems to support the management of an important aquaculture species, Manila clam (Ruditapes philippinarum). We use the Venice lagoon (Italy), the first site in Europe for the production of R. philippinarum, to illustrate the potential of this modelling approach. To investigate the relationship between the yield of R. philippinarum and a set of environmental factors, we used a Random Forest (RF) algorithm. The RF model was tuned with a large data set (n = 1698) and validated by an independent data set (n = 841). Overall, the model provided good predictions of site-specific yields and the analysis of marginal effect of predictors showed substantial agreement among the modelled responses and available ecological knowledge for R. philippinarum. The most influent environmental factors for yield estimation were percentage of sand in the sediment, salinity, and water depth. Our results agree with findings from other North Adriatic lagoons. The application of the fitted RF model to continuous maps of all the environmental variables allowed estimates of the potential yield for the whole basin. Such a spatial representation enabled site-specific estimates of yield in different farming areas within the lagoon. We present a possible management application of our model by estimating the potential yield under the current farming distribution and comparing it to a proposed re-organization of the farming areas. Our analysis suggests a reduction of total yield is likely to result from the proposed re-organization.     

Incorporating low-resolution historic species location data decreases performance of distribution models

Developing robust species distribution models is important as model outputs are increasingly being incorporated into conservation policy and management decisions. A largely overlooked component of model assessment and refinement is whether to include historic species occurrence data in distribution models to increase the data sample size. Data of different temporal provenance often differ in spatial accuracy and precision. We test the effect of inclusion of historic coarse-resolution occurrence data on distribution model outputs for 187 species of birds in Australian tropical savannas. Models using only recent (after 1990), fine-resolution data had significantly higher model performance scores measured with area under the receiver operating characteristic curve (AUC) than models incorporating both fine- and coarse-resolution data. The drop in AUC score is positively correlated with the total area predicted to be suitable for the species (R² = 0.163-0.187, depending on the environmental predictors in the model), as coarser data generally leads to greater predicted areas. The remaining unexplained variation is likely to be due to the covariate errors resulting from resolution mismatch between species records and environmental predictors. We conclude that decisions regarding data use in species distribution models must be conscious of the variation in predictions that mixed-scale datasets might cause.     

Increasing accuracy of dispersal kernels in grid-based population models

Dispersal kernels in grid-based population models specify the proportion, distance and direction of movements within the model landscape. Spatial errors in dispersal kernels can have large compounding effects on model accuracy. Circular Gaussian and Laplacian dispersal kernels at a range of spatial resolutions were investigated, and methods for minimizing errors caused by the discretizing process were explored. Kernels of progressively smaller sizes relative to the landscape grid size were calculated using cell-integration and cell-center methods. These kernels were convolved repeatedly, and the final distribution was compared with a reference analytical solution. For large Gaussian kernels (σ > 10 cells), the total kernel error was <10⁻¹¹ compared to analytical results. Using an invasion model that tracked the time a population took to reach a defined goal, the discrete model results were comparable to the analytical reference. With Gaussian kernels that had σ ≤ 0.12 using the cell integration method, or σ ≤ 0.22 using the cell center method, the kernel error was greater than 10%, which resulted in invasion times that were orders of magnitude different than theoretical results. A goal-seeking routine was developed to adjust the kernels to minimize overall error. With this, corrections for small kernels were found that decreased overall kernel error to <10⁻¹¹ and invasion time error to <5%.     

Influence of grain size on species-habitat models

High resolution remote sensing data facilitate the use of small-scale habitat features such as trees or hedges in the analysis of species-habitat relationships. Such data potentially enable more accurate species-habitat mapping than lower resolution data. Here, for the first time, we systematically investigated this hypothesis by altering the spatial resolution from 1 m up to 1000 m grain size in species-habitat models of 13 bird species. The study area covered the Nidda river catchment in central Germany, a large heterogeneous landscape of 1620 km². A high resolution habitat map of the area was converted to coarser spatial and thematic resolutions in seven steps. We investigated how model performance responded to grain size, and we compared the differential effects of spatial resolution and thematic resolution on model performance. Explained deviance (D²) of the bird models generally decreased with coarser spatial resolution of the data, although it did not decrease monotonically in all species. On average across all species, model D² decreased from 41.5 at 1 m grain size to 15.9 at 1000 m grain size. Ten species were best modelled at 1 m, two species at 3 m and one species at 32 m grain size. Model performance degraded continuously with increasing grain size, both in habitat generalist and habitat specialist bird species, and was systematically lower in habitat generalists. The higher model performance observed at finer grain sizes was most likely caused by the combination of three factors: (1) high spatial accuracy of bird records and (2) a more precise location and delineation of habitat features and, (3) to a lesser degree, by more habitat types differentiated in maps of finer resolution. We conclude that higher spatial and thematic resolution data can be essential for deriving accurate predictions on bird distribution patterns from species-habitat models. Especially for bird species that are sensitive to specific land-use types or to small-scaled habitat features, a grain size of 1-3 m seems most promising.     

Optimization of ecosystem model parameters using spatio-temporal soil moisture information

Parameters in process-based terrestrial ecosystem models are often nonlinearly related to the water flux to the atmosphere, and they also change temporally and spatially. Therefore, for estimating soil moisture, process-based terrestrial ecosystem models inevitably need to specify spatially and temporally variant model parameters. This study presents a two-stage data assimilation scheme (TSDA) to spatially and temporally optimize some key parameters of an ecosystem model which are closely related to soil moisture. At the first stage, a simplified ecosystem model, namely the Boreal Ecosystem Productivity Simulator (BEPS), is used to obtain the prior estimation of daily soil moisture. After the spatial distribution of 0–10 cm surface soil moisture is derived from remote sensing, an Ensemble Kalman Filter is used to minimize the difference between the remote sensing model results, through optimizing some model parameters spatially. At the second stage, BEPS is reinitialized using the optimized parameters to provide the updated model predictions of daily soil moisture. TSDA has been applied to an arid and semi-arid area of northwest China, and the performance of the model for estimating daily 0–10 cm soil moisture after parameter optimization was validated using field measurements. Results indicate that the TSDA developed in this study is robust and efficient in both temporal and spatial model parameter optimization. After performing the optimization, the correlation (r²) between model-predicted 0–10 cm soil moisture and field measurement increased from 0.66 to 0.75. It is demonstrated that spatial and temporal optimization of ecosystem model parameters can not only improve the model prediction of daily soil moisture but also help to understand the spatial and temporal variation of some key parameters in an ecosystem model and the corresponding ecological mechanisms controlling the variation.     

Development of TRIPLEX-Management model for simulating the response of forest growth to pre-commercial thinning

In order to simulate forest growth response to pre-commercial thinning (PCT), TRIPLEX1.0 - a process-based model designed to predict forest growth as well as carbon (C) and nitrogen (N) dynamics - was modified and improved to also simulate managed forest ecosystem thinning practices. A three-parameter Weibull distribution model was integrated to simulate thinning treatments within the newly developed TRIPLEX-Management model. The thinning intensity component within the model allows users to simulate thinning treatments by applying basal area, stand density and volume to quantify thinning intensity. Natural mortality decreased following thinning due to an increase in growing space for residual stems. Predicted litterfall pools also increased after thinning events took place. The TRIPLEX-Management model was tested against published observational data for Jack Pine (Pinus banksiana Lamb.) stands subjected to PCT in Northwestern Ontario, Canada. The coefficients of determination (R²) between the predicted and observed variables including stand density, mean DBH (diameter at breast height), the quadratic mean DBH, total volume and merchantable volume as well as belowground, aboveground, and total biomass ranged from 0.50 to 0.88 (n = 20, P < 0.001) with the exception of mean tree height (R² = 0.25, n = 20, P < 0.05). Overall, the Willmott index of agreement between predicted and observed variables ranged from 0.97 to 1.00. Results show that the TRIPLEX-Management model is generally capable of simulating growth response to PCT for Jack Pine stands.     

Simulating soil organic matter with CQESTR (v. 2.0): Model description and validation against long-term experiments across North America

Soil carbon (C) models are important tools for examining complex interactions between climate, crop and soil management practices, and to evaluate the long-term effects of management practices on C-storage potential in soils. CQESTR is a process-based carbon balance model that relates crop residue additions and crop and soil management to soil organic matter (SOM) accretion or loss. This model was developed for national use in U.S and calibrated initially in the Pacific Northwest. Our objectives were: (i) to revise the model, making it more applicable for wider geographic areas including potential international application, by modifying the thermal effect and incorporating soil texture and drainage effects, and (ii) to recalibrate and validate it for an extended range of soil properties and climate conditions. The current version of CQESTR (v. 2.0) is presented with the algorithms necessary to simulate SOM at field scale. Input data for SOM calculation include crop rotation, aboveground and belowground biomass additions, tillage, weather, and the nitrogen content of crop residues and any organic amendments. The model was validated with long-term data from across North America. Regression analysis of 306 pairs of predicted and measured SOM data under diverse climate, soil texture and drainage classes, and agronomic practices at 13 agricultural sites having a range of SOM (7.3–57.9 g SOM kg⁻¹), resulted in a linear relationship with an r² of 0.95 (P < 0.0001) and a 95% confidence interval of 4.3 g SOM kg⁻¹. Using the same data the version 1.0 of CQESTR had an r² of 0.71 with a 95% confidence interval of 5.5 g SOM kg⁻¹. The model can be used as a tool to predict and evaluate SOM changes from various management practices and offers the potential to estimate C accretion required for C credits.     

Estimation of annual potential evapotranspiration at regional scale based on the effect of moisture on soil respiration

Potential evapotranspiration (PET) is an important component of water cycle. For traditional models derived from the principle of aerodynamics and the surface energy balance, its calculation always includes many parameters, such as net radiation, water vapor pressure, air temperature and wind speed. We found that it can be acquired in an easier way in specific regions. In this study, a new PET model (PETP model) derived from two empirical models of soil respiration was evaluated using the Penman-Monteith equation as a standard method. The results indicate that the PETP model estimation concur with the Penman-Monteith equation in sites where annual precipitation ranges from 717.71 mm to 1727.37 mm (R² = 0.68, p = 0.0002), but show large discrepancies in all sites (R² = 0.07, p = 0.1280). Then we applied our PETP model at the global scale to the regions with precipitation higher than 700 mm using 2.5° CMAP data to obtain the annual PET for 2006. As expected, the spatial pattern is satisfactory overall, with the highest PET values distributed in the lower latitudes or coastal regions, and with an average of 1292.60 ± 540.15 mm year⁻¹. This PETP model provides a convenient approach to estimate PET at regional scales.     

River network structure shapes interannual feedbacks between adult sea lamprey migration and larval habitation

Adult sea lampreys (Petromyzon marinus) migrating upstream to spawn follow a pheromone released by instream larvae. The size (i.e. flow) of a tributary dilutes the concentration of this pheromone, such that the downstream propagation pattern of larval pheromone must be influenced by patterns in the relative sizes and numbers of confluent tributaries. We developed an individual-based model to explicitly test the resulting hypothesis that river network structure influences the migration decisions of adult lampreys following the larval pheromone, and in turn the distribution of larvae. First, we initialized the model using randomly generated river networks, and found a strong positive relationship between network diameter and larval aggregation. Larvae aggregated over time, and the degree and rate of this aggregation depended on network diameter. Second, we initialized the model using a river network based on the Muskegon River, Michigan, and compared model-generated larval distribution to available field survey data. We found a significant correlation between model-generated larval abundance and field-measured larval densities (r₂ = 0.54; p < 0.0001). We also found an inverse relationship between subwatershed area and the degree to which path-dependent effects influenced larval abundance in that subwatershed. Our results overall suggest that larval distribution across a watershed results from a system of context-dependent interannual feedbacks shaped by network structure and the past migratory and spawning behavior of adults.     

Modeling a spatially restricted distribution in the Neotropics: How the size of calibration area affects the performance of five presence-only methods

We here examine species distribution models for a Neotropical anuran restricted to ombrophilous areas in the Brazilian Atlantic Forest hotspot. We extend the known occurrence for the treefrog Hypsiboas bischoffi (Anura: Hylidae) through GPS field surveys and use five modeling methods (BIOCLIM, DOMAIN, OM-GARP, SVM, and MAXENT) and selected bioclimatic and topographic variables to model the species distribution. Models were first trained using two calibration areas: the Brazilian Atlantic Forest (BAF) and the whole of South America (SA). All modeling methods showed good levels of predictive power and accuracy with mean AUC ranging from 0.77 (BIOCLIM/BAF) to 0.99 (MAXENT/SA). MAXENT and SVM were the most accurate presence-only methods among those tested here. All but the SVM models calibrated with SA predicted larger distribution areas when compared to models calibrated in BAF. OM-GARP dramatically overpredicted the species distribution for the model calibrated in SA, with a predicted area around 10⁶ km² larger than predicted by other SDMs. With increased calibration area (and environmental space), OM-GARP predictions followed changes in the environmental space associated with the increased calibration area, while MAXENT models were more consistent across calibration areas. MAXENT was the only method that retrieved consistent predictions across calibration areas, while allowing for some overprediction, a result that may be relevant for modeling the distribution of other spatially restricted organisms.     

Assessing spatial uncertainty of the Portuguese fire risk through direct sequential simulation

Forest fires have a significant economic, social, and environmental impact in Portugal. For that its fire risk was assessed through Bayes Formalism, where the main component of the risk of fire was assessed by the conditional probability of fire I(u,t) given a class of the daily severity rating (DSR) for a specific period of time—P[I(u,t)|R(u,t)]. The evaluation of this a posterior probability, P[I(u,t)|R(u,t)], was based on the update of marginal local probability of fire in each chosen region u (Durão, 2006).DSR values were used to calculate fire's risk, taking into account historical data, I(s,t), in a given region s, and also to define DSR's local thresholds in order to have P [I(u,t)|R(u,t)] ≥ 0.65.In this paper we characterize these posterior probabilities using direct sequential simulation models (DSS models) to obtain the spatial distribution of these probabilities over the entire Portugal, in order to assess the risk of fire and associated spatial uncertainty. Local probability density functions (pdfs) and spatial uncertainty are evaluated by a set of equiprobable simulated images of these posterior probabilities.Results are presented and discussed for the Portuguese fire seasons of the 2-year period, 2003-2004. The conditional probabilities reproduced reasonably well what was officially published for the studied fire seasons. We expect that a better understanding of both spatial and temporal patterns of fire in Portugal together with uncertainty measures constitutes an important tool for managers, helping to improve the effectiveness of fire prevention, detection and fire fighting resources allocation in critical social and environmental areas.     

Residual levels and health risk of polycyclic aromatic hydrocarbons in freshwater fishes from Lake Small Bai-Yang-Dian, Northern China

The residual levels of polycyclic aromatic hydrocarbons (PAHs) in the liver, brain, gill and muscle tissues of four common edible freshwater fish species including crucian carp, snakehead fish, grass carp and silver carp collected from Lake Small Bai-Yang-Dian in northern China were measured by GC-MS. The distribution and composition pattern of PAHs in the fish tissues, and the effects of lipid contents in fish tissues and the octanol-water partition coefficient (Kow) of PAHs congeners on them were analyzed. The human health risk of PAHs though fish consumptions was estimated. The following results were obtained: (1) The average residual levels of total PAHs (PAH₁₆) on wet weight base in the different tissues of each fish species ranged from 4.764 to 144.254 ng/g ww. The differences in the average residual levels on wet weight base for PAH₁₆ within four fish species were not statistically significant (P > 0.05); however, these within four fish tissues were statistically significant (P < 0.01). (2) There were very similar distribution patterns of PAH congeners among both the fish tissues and the fish species, as indicated by statistically significant positive interrelationships (R = 0.58-0.97, P < 0.01 or P < 0.05). Low molecular weight (LMW) PAHs predominated the distribution in the fish tissues, accounting for 89.97% of total PAHs. Phe was the most dominant component, according for 37.79% of total PAHs, followed by Ant (18.59%), Flo (12.59%), Nap (10.79%), Fla (9.82%) and Pyr (6.43%). (3) The PAHs residues and distribution in the fish tissues are dependent on both the Kow of PAH congeners and the lipid contents in the fish tissues. There was a significant positive relationship (R = 0.7116, P < 0.0001) between lipid contents and PAHs residual levels. The statistically significant negative relationships (P < 0.05) were found between LogKow and log-transformed PAHs contents on wet weight base for all fish tissues except for the muscle tissue of snakehead fish, the brain and liver tissues of crucian carp. (4) The risk levels of total PAHs were lower than 10⁻⁵ for the muscle tissues of four studied fish species and for the brain tissues of grass carp and snakehead fish; while these were higher than 10⁻⁵ for the brain tissues of crucian carp and silver carp. The risk levels of total PAHs in the liver tissues of four studied fish species except for snakehead fish exceeded 10⁻⁵ for 2-4.5 times. However, the potency equivalent concentration (PEC) of total PAHs in four studied fish tissues were still lower than the maximum permissible BaP limits for crops and baked meat and for plants in the national criterions. The distributions of PAH congeners in fish were well simulated by a level III fugacity model, especially for low molecule weight PAHs.     

Modelling habitat overlap among sympatric mesocarnivores in southern Illinois,USA

Few researchers have developed large-scale habitat models for sympatric carnivore species. We created habitat models for red foxes (Vulpes vulpes), coyotes (Canis latrans) and bobcats (Lynx rufus) in southern Illinois, USA, using the Penrose distance statistic, remotely sensed landscape data, and sighting location data within a GIS. Our objectives were to quantify and spatially model potential habitat differences among species. Habitat variables were quantified for 1-km² buffered areas around mesocarnivore sighting locations. Following variable reduction procedures, five habitat variables (percentage of grassland patches, interspersion–juxtaposition of forest patches, mean fractal dimension of wetland patches and the landscape, and road density) were used for analysis. Only one variable differed (P < 0.05) between red fox and coyote sighting areas (road density) and bobcat and coyote sighting areas (mean fractal dimension of the landscape). However, all five variables differed between red fox and bobcat sighting areas, indicating considerable differences in habitat affiliation between this pair-group. Compared to bobcats, red fox sightings were affiliated with more grassland cover and larger grassland patches, higher road densities, lower interspersion and juxtaposition of forest patches, and lower mean fractal dimension of wetland patches. These differences can be explained by different life history requirements relative to specific cover types. We then used the Penrose distance statistic to create habitat models for red foxes and bobcats, respectively, based on the five-variable dataset. An independent set of sighting locations were used to validate these models; model fit was good with 65% of mesocarnivore locations within the top 50% of Penrose distance values. In general, red foxes were affiliated with mixtures of agricultural and grassland cover, whereas bobcats were associated with a combination of grassland, wetland, and forest cover. The greatest habitat overlap between red foxes and bobcats was found at the interface between forested areas and more open cover types. Our study provides insight into habitat overlap among sympatric mesocarnivores, and the distance-based modelling approach we used has numerous applications for modelling wildlife–habitat relationships over large scales.     

Individual-based modeling of flooding and salinity effects on a coastal swamp forest

Coastal swamps are among the rapidly vanishing wetland habitats in Louisiana. Increased flooding, nutrient and sediment deprivation, and salt-water intrusion have been implicated as probable causes of the decline of coastal swamps. We developed a two-species individual-based forest succession model to compare the growth and composition of a cypress-tupelo swamp under various combinations of flooding intensity and salinity levels, using historical time-series of stage and salinity data as inputs. Our model simulates forest succession over 500 years by representing the growth, mortality, and reproduction of individual Taxodium distichum (baldcypress) and Nyssa aquatica (water tupelo) trees in a 1-km² spatial grid of 10 m × 10 m cells that vary in water levels and salinity through differences in elevation. We independently adjusted the elevations of each cell to obtain different grid-wide mean elevations and standard deviations of elevation; this affected the temporal and spatial pattern of flooding. We calibrated the model by adjusting selected parameters until averaged basal area, stem density and wood production rates under two different mean elevations (partially versus highly flooded) were qualitatively similar to comparable values reported for swamps in the literature. Corroboration involved comparing model predictions to four well-monitored contrasting habitat sites within the Maurepas Basin, Louisiana, USA. Model predictions of both species combined showed the same patterns among sites as the data, but the model overestimated wood production and the dominance of T. distichum. Exploratory simulations predicted that increased flooding leads to swamps with reduced basal areas and stem densities, while increased salinity resulted in lower basal areas at low salinity concentration (∼1-3 psu) and complete tree mortality at higher salinity concentrations (∼2-6 psu). Our model can provide insight into the succession dynamics of coastal swamps and information for the effective design of restoration actions.     

Generality of Models that Predict the Distribution of Species: Conservation Activity and Reduction of Model Transferability for a Threatened Bustard

Abstract:  Predictive models can help clarify the distribution of poorly known species but should display strong transferability when applied to independent data. Nevertheless, model transferability for threatened tropical species is poorly studied. We built models predicting the incidence of the critically endangered Bengal Florican ( Houbaropsis bengalensis ) within the Tonle Sap (TLS) floodplain, Cambodia. Separate models were constructed with soil, land-use, and landscape data and species incidence sampled over the entire floodplain (12,000 km²) and from the Kompong Thom (KT) province (4000 km²). In each case, the probability of Bengal Florican presence within randomly selected 1 × 1 km squares was modeled by binary logistic regression with multimodel inference. We assessed the transferability of the KT model by comparing predictions with observed incidence elsewhere in the floodplain. In terms of standard model-validation statistics, the KT model showed good spatial transferability. Nevertheless, it overpredicted florican presence outside the KT calibration region, classifying 491 km² as suitable habitat compared with 237 km² predicted as suitable by the TLS model. This resulted from higher species incidence within the calibration region, probably owing to a program of conservation education and enforcement that has reduced persecution there. Because both research and conservation activity frequently focus on areas with higher density, such effects could be widespread, reducing transferability of predictive distribution models.     

Inter-annual variability of ecosystem production in boreal jack pine forests (1975-2004) estimated from tree-ring data using CBM-CFS3

We describe and apply a method of using tree-ring data and an ecosystem model to reconstruct past annual rates of ecosystem production. Annual data on merchantable wood volume increment and mortality obtained by dendrochronological stand reconstruction were used as input to the Carbon Budget Model of the Canadian Forest Sector (CBM-CFS3) to estimate net ecosystem production (NEP), net primary production (NPP), and heterotrophic respiration (Rh) annually from 1975 to 2004 at 10 boreal jack pine (Pinus banksiana Lamb.) stands in Saskatchewan and Manitoba, Canada. From 1975 (when sites aged 41-60 years) to 2004 (when they aged 70-89 years), all sites were moderate C sinks except during some warmer than average years where estimated Rh increased. Across all sites and years, estimated annual NEP averaged 57 g Cm⁻² yr⁻¹ (range −31 to 176 g Cm⁻² yr⁻¹), NPP 244 g Cm⁻² yr⁻¹ (147-376 g Cm⁻² yr⁻¹), and Rh 187 g Cm⁻² yr⁻¹ (124-270 g Cm⁻² yr⁻¹). Across all sites, NPP was related to stand age and density, which are proxies for successional changes in leaf area. Regionally, warm spring temperature increased NPP and defoliation by jack pine budworm 1 year previously reduced NPP. Our estimates of NPP, Rh, and NEP were plausible when compared to regional eddy covariance and carbon stock measurements. Inter-annual variability in ecosystem productivity contributes uncertainty to inventory-based assessments of regional forest C budgets that use yield curves predicting averaged growth over time. Our method could expand the spatial and temporal coverage of annual forest productivity estimates, providing additional data for the development of empirical models accounting for factors not presently considered by these models.