Ecological Datasets Research Articles

The EcoData retriever: improving access to existing ecological data.

Ecological research relies increasingly on the use of previously collected data. Use of existing datasets allows questions to be addressed more quickly, more generally, and at larger scales than would otherwise be possible. As a result of large-scale data collection efforts, and an increasing emphasis on data publication by journals and funding agencies, a large and ever-increasing amount of ecological data is now publicly available via the internet. Most ecological datasets do not adhere to any agreed-upon standards in format, data structure or method of access. Some may be broken up across multiple files, stored in compressed archives, and violate basic principles of data structure. As a result acquiring and utilizing available datasets can be a time consuming and error prone process. The EcoData Retriever is an extensible software framework which automates the tasks of discovering, downloading, and reformatting ecological data files for storage in a local data file or relational database. The automation of these tasks saves significant time for researchers and substantially reduces the likelihood of errors resulting from manual data manipulation and unfamiliarity with the complexities of individual datasets.

Ecohydrological Responses to Diversion of Groundwater: Case Study of a Deep-Rock Repository for Spent Nuclear Fuel in Sweden

Planning and license applications concerning groundwater diversion in areas containing water-dependent or water-favored habitats must take into account both hydrological effects and associated ecological consequences. There is at present no established methodology to assess such ecohydrological responses. Thus, this paper describes a new stepwise methodology to assess ecohydrological responses to groundwater diversion from, e.g., water-drained pits, shafts, tunnels, and caverns in rock below the groundwater table. The methodology is illustrated using the planned deep-rock repository for spent nuclear fuel at Forsmark in central Sweden as a case study, offering access to a unique hydrological and ecological dataset. The case study demonstrates that results of ecohydrological assessments can provide useful inputs to planning of monitoring programs and mitigation measures in infrastructure projects. As a result of the assessment, artificial water supply to wetlands is planned in order to preserve biological diversity, nature values, and vulnerable species.Electronic supplementary materialThe online version of this article (doi:10.1007/s13280-013-0404-8) contains supplementary material, which is available to authorized users.

A general framework for analyzing beta diversity, nestedness and related community-level phenomena based on abundance data

We describe a procedure for evaluating the relative importance of beta diversity, nestedness, and similarity properties of ecological data matrices containing density, cover or biomass scores of species. Our goals are achieved by extension of the simplex approach – originally proposed for presence–absence data – to abundances. Basically, the method involves decomposition of the Marczewski–Steinhaus coefficient of dissimilarity between pairs of sites into two fractions, one derived from differences between total abundance and the other from differences due to abundance replacement. These are contrasted by the similarity function counterpart, known as the Ruzicka coefficient, and are displayed graphically using ternary (or 2D simplex) plots. Interpretation is aided by calculating percentage contributions from these components to the (dis)similarity structure. Measures of replacement and nestedness are new for abundance data; these are considered complementary phenomena reflecting antithetic ecological processes that are analogous to those operating at the presence–absence level. The method is illustrated by artificial data and a range of actual ecological data sets representing different groups of organisms, different scales and different types of data. While the simplex diagrams and associated coefficients are meaningful by themselves, their comparison with presence–absence based results gives additional insight into data structure and background factors.

Spherical k‐means clustering is good for interpreting multivariate species occurrence data

Summary Clustering multivariate species data can be an effective way of showing groups of species or samples with similar characteristics. Most current techniques classify the samples first and then the species. A disadvantage of classifying the samples first is that relatively subtle differences between occurrence profiles of species can be obscured. The k‐means method of clustering minimizes the sum of squared distances between cluster centres and cluster members. If the entities to be clustered are projected on the unit sphere, then a natural measure of dispersion is the sum of squared chord distances separating the entities from their cluster centres; k‐means clustering with this measure of dispersion is called spherical k‐means (SKM). We also consider a variant in which the sum of squared perpendicular distances to a central ray is minimized. Unweighted SKM is liable to produce clusters of very rare species. This feature can be avoided if each point on the unit sphere is weighted by the length of the ray that produced it. The standard SKM algorithm converges to very numerous local optima. To avoid this problem, we have developed a computationally intensive algorithm that uses multiple randomizations to select high‐quality seed species. The species clustering can be used to define simplified attributes for the samples. If the samples are then classified using the same technique, the resulting matrix of clustered species and clustered samples provides a biclustering of the data. The strength of the relationship between clusters can be measured by their mutual information, which is effectively the entropy of the biclustering. The technique was tested on five ecological and biogeographical datasets ranging in size from 30 species in 20 samples to 1405 species in 3857 samples. Several variants of SKM were compared, together with results from the established program Twinspan. When judged by entropy, SKM always performed adequately and produced the best clustering in all datasets but the smallest.

Eye-Size Variability in Deep-Sea Lanternfishes (Myctophidae): An Ecological and Phylogenetic Study

One of the most common visual adaptations seen in the mesopelagic zone (200–1000 m), where the amount of light diminishes exponentially with depth and where bioluminescent organisms predominate, is the enlargement of the eye and pupil area. However, it remains unclear how eye size is influenced by depth, other environmental conditions and phylogeny. In this study, we determine the factors influencing variability in eye size and assess whether this variability is explained by ecological differences in habitat and lifestyle within a family of mesopelagic fishes characterized by broad intra- and interspecific variance in depth range and luminous patterns. We focus our study on the lanternfish family (Myctophidae) and hypothesise that lanternfishes with a deeper distribution and/or a reduction of bioluminescent emissions have smaller eyes and that ecological factors rather than phylogenetic relationships will drive the evolution of the visual system. Eye diameter and standard length were measured in 237 individuals from 61 species of lanternfishes representing all the recognised tribes within the family in addition to compiling an ecological dataset including depth distribution during night and day and the location and sexual dimorphism of luminous organs. Hypotheses were tested by investigating the relationship between the relative size of the eye (corrected for body size) and variations in depth and/or patterns of luminous-organs using phylogenetic comparative analyses. Results show a great variability in relative eye size within the Myctophidae at all taxonomic levels (from subfamily to genus), suggesting that this character may have evolved several times. However, variability in eye size within the family could not be explained by any of our ecological variables (bioluminescence and depth patterns), and appears to be driven solely by phylogenetic relationships.

Designing a sustainable land use scenario based on a combination of ecological assessments and economic optimization

Land use change and the corresponding effects on ecosystems and their services has gained much interest in the recent past, particularly in areas with a significant reservoir of biodiversity, the so-called biodiversity hot spots. In order to assess the impact of possible future land use decisions in a watershed in Yunnan, Southwest China, we applied a method of combining ecological and socio-economic indicators to highlight key aspects concerning the current status of our research area. Data on species diversity, landscape matrix and erosion risk as well as agricultural and socio-economic activities were gathered and analyzed. We were able to locate the areas were conservation measures, erosion control and improved agricultural practices would have the strongest impacts. This information was used to develop a storyline for a “Go Green” scenario. Expert groups and interdisciplinary panels were used to critically review, enhance and expand this storyline in the area of conflict between nature conservation, rural livelihood and economic development.Based on the set of planning prerequisites, a village-household linear programming model was developed and solved with the General Algebraic Modelling System (GAMS) to identify factors driving landscape and land use changes for three different farming systems in the Naban River Watershed National Nature Reserve, mainly to contribute to the CLUE_Naban model by providing representative farm types and to analyze the decision making of land use (until 2025). In addition, this model is designed to provide policy makers with potential strategic intervention options for land use planning through the utilization of shadow prices.This process enabled us to reconcile the demands for nature conservation and economic wellbeing on a basis of an iterative and participatory working process that incorporates ecological and economic datasets, but also takes the sustainability of rural livelihood into account.

Species Assemblage Comparison with Abundance-based Data Using Zero-inflated Poisson Mixtures

Species assemblage comparison is important in many ecological studies. In this paper, we develop a novel test for comparing species assemblages when abundance data from multiple quadrats are available. The test is based on the zero-inflated Poisson mixture model which we introduce to characterize the species assemblage given abundance data from multiple quadrats. We present a simulation study to evaluate the performance of our proposed test. The application of our test is further demonstrated on an ecological dataset. Comparison of species assemblages has important applications in ecology, since it provides crucial information about the spatial and temporal variations of ecosystems. There are two typical types of data collected in ecological studies: abundance-based data that contains the information of counts of each observed species in each sampling unit, and incidence-based data that only notes whether a species is present or absent in each sampling unit. Depending on the sampling procedure, the abundance-based data can be further divided into two categories. In one, the whole sampling area is treated as a single sampling unit, and the count information of each observed species is summarized for the whole area. The other has the sampling area divided into numerous plots, a sample of plots is randomly taken, and the count information of each observed species is recorded for each of the sampled plots. Following the terminology commonly used in ecology, we call those plots quadrats. We refer to the first type as abundance data from a single quadrat, and refer to the second type as abundance data from multiple quadrats. In the literature, mixture models are popular choices to model the ecological data due to their capabilities to account for heterogeneity among species (see, for example, Ord and Whitmore (1986), Bunge and Fitzpatrick (1993), Chao and Bunge (2002), Bohning and Schon (2005), Mao and Colwell (2005), Mao (2006)). More specifically, for incidence-based data, the binomial mixture model is usually

Emulator-assisted reduced-rank ecological data assimilation for nonlinear multivariate dynamical spatio-temporal processes

As ecological data sets increase in spatial and temporal extent with the advent of new remote sensing platforms and long-term monitoring networks, there is increasing interest in forecasting ecological processes. Such forecasts require realistic initial conditions over complete spatial domains. Typically, data sources are incomplete in space, and the processes include complicated dynamical interactions across physical and biological variables. This suggests that data assimilation, whereby observations are fused with mechanistic models, is the most appropriate means of generating complete initial conditions. Often, the mechanistic models used for these procedures are very expensive computationally. We demonstrate a rank-reduced approach for ecological data assimilation whereby the mechanistic model is based on a statistical emulator. Critically, the rank-reduction and emulator construction are linked and, by utilizing a hierarchical framework, uncertainty associated with the dynamical emulator can be accounted for. This provides a so-called “weak-constraint” data assimilation procedure. This approach is demonstrated on a high-dimensional multivariate coupled biogeochemical ocean process.

The importance of prior choice in model selection: a density dependence example

Summary It is important to discern the magnitude of density dependence a species exhibits, as well as the time lag over which it operates. Knowledge of a species' likely response to natural as well as synthetic shocks will assist in effective species management. Statistically this is a challenging problem which does not usually admit closed‐form mathematical analysis. Consequently, many people have used Bayesian methods to fit state space models of density dependence to many different species, of which we take eleven species of North American duck as our motivating examples. A Bayesian analysis requires a choice of model and parameter prior. The latter is difficult to do without inducing bias in model selection, and we attempt to address this problem. Our priors will be obtained by considering which parameter values are representative of features we expect to see in the data, and which would produce unnatural behaviour. To fit the models, we use a novel sequential Monte Carlo method (particle learning) not previously applied to ecological data sets. We show that existing analyses on the duck data may have been susceptible to a common problem in Bayesian model selection (Lindley's paradox), and suggest methods for prior selection which mitigate this issue. We also discover that although it is possible to detect the existence of density dependence, it is unrealistic to expect to determine the time lag over which it operates without a great deal of data, even if said data are simulated from the model. We demonstrate that prior choices motivated by the above considerations can lead to substantially increased predictive accuracy over surprisingly long time scales whether model selection is of primary concern or not. We conclude from our analysis of real‐world data that there is little evidence of density dependence in many duck species, suggesting that such effects, if present, are likely to be small in magnitude.

Development and assessment of ecological models in the context of the European Water Framework Directive: Key issues for trainers in data-driven modeling approaches

Teaching students to develop data-driven models is a challenging task as a good balance has to be found between the theoretical background of the models, the ecological relevance of the knowledge rules inferred and their socio-economic applicability. In this context it is unclear which aspects of the modeling process are easily understood by students, and in particular, how theoretical issues interfere with practical boundary conditions and socio-economic relevance (ecosystem protection, water management, policy development, ecological engineering). In order to fill this knowledge gap, students developed static data-driven models and tutors assessed students' performances. Criteria such as the theoretical, ecological and socio-economic relevance of the derived knowledge rules were used to select the most optimal models.We noticed an inverse relationship between the complexity of the subtasks and the number of students that succeeded. Students evaluated their models with respect to the theoretical reliability, but were not likely to consider the other two criteria. Half of the students succeeded in assessing the models based on their ecological relevance and only 17% of the students checked the socio-economic relevance of the knowledge rules. Four groups out of seven assessed their models merely based on the predictive power of the models. Only one group integrated the theoretical, ecological and socio-economic relevance to assess the models.The key findings of our research can be used to optimize the efficiency of data mining courses. We reveal which aspects of the modeling process students seem to overemphasize and give recommendations about the topics trainers should emphasize in the future to ensure that students develop advanced skills. Based on our results the theory–practice dichotomy in higher education can be further reduced. Our learning-by-doing approach showed students how to solve common problems in ecological data sets (e.g. missing data, outliers, collinearity, non-normal distribution, parameterization, uncertainty, etc.), which are often only briefly discussed in basic statistical courses.

A spectral and Bayesian approach for analysis of fluctuations and synchrony in ecological datasets

Summary Autocorrelation within ecological time series and synchrony between them may provide insight into the main drivers of observed dynamics. We here present methods that analyse autocorrelation and synchrony in ecological datasets using a spectral approach combined with Bayesian inference. To exemplify, we implement the method on dendrochronological data of the pedunculate oak (Quercus robur). The data consist of 110 years of growth of 10 live trees and seven trees that died during a synchronized oak death in Sweden in c. 2002–2007. We find that the highest posterior density is found for a noise colour of tree growth of γ ≈ 0·95 (i.e. ‘pink noise’) with little difference between trees, suggesting climatic variation as a driving factor. This is further supported by the presence of synchrony, which we estimate based on phase‐shift analysis. We conclude that the synchrony is time‐scale dependent with higher synchrony at larger time‐scales. We further show that there is no difference between the growth patterns of the alive and dead tree groups. This suggests that the trees were driven by the same factors prior to the synchronized death. We argue that this method is a promising approach for linking theoretical models with empirical data.

Integrating Paleobiology, Archeology, and History to Inform Biological Conservation

The search for novel approaches to establishing ecological baselines (reference conditions) is constrained by the fact that most ecological studies span the past few decades, at most, and investigate ecosystems that have been substantially altered by human activities for decades, centuries, or more. Paleobiology, archeology, and history provide historical ecological context for biological conservation, remediation, and restoration. We argue that linking historical ecology explicitly with conservation can help unify related disciplines of conservation paleobiology, conservation archeobiology, and environmental history. Differences in the spatial and temporal resolution and extent (scale) of prehistoric, historic, and modern ecological data remain obstacles to integrating historical ecology and conservation biology, but the prolonged temporal extents of historical ecological data can help establish more complete baselines for restoration, document a historical range of ecological variability, and assist in determining desired future conditions. We used the eastern oyster (Crassostrea virginica) fishery of the Chesapeake Bay (U.S.A.) to demonstrate the utility of historical ecological data for elucidating oyster conservation and the need for an approach to conservation that transcends disciplinary boundaries. Historical ecological studies from the Chesapeake have documented dramatic declines (as much as 99%) in oyster abundance since the early to mid-1800 s, changes in oyster size in response to different nutrient levels from the sixteenth to nineteenth centuries, and substantial reductions in oyster accretion rates (from 10 mm/year to effectively 0 mm/year) from the Late Holocene to modern times. Better integration of different historical ecological data sets and increased collaboration between paleobiologists, geologists, archeologists, environmental historians, and ecologists to create standardized research designs and methodologies will help unify prehistoric, historic, and modern time perspectives on biological conservation.

The WISER metadatabase: the key to more than 100 ecological datasets from European rivers, lakes and coastal waters

In ecological sciences, the role of metadata (i.e. key information about a dataset) to make existing datasets visible and discoverable has become increasingly important. Within the EU-funded WISER project (Water bodies in Europe: Integrative Systems to assess Ecological status and Recovery), we designed a metadatabase to allow scientists to find the optimal data for their analyses. An online questionnaire helped to collect metadata from the data providers and an online query tool (http://www.wiser.eu/results/meta-database/) facilitated data evaluation. The WISER metadatabase currently holds information on 114 datasets (22 river, 71 lake, 1 general freshwater and 20 coastal/transitional datasets), which also can be accessed by external scientists. We evaluate if generally used metadata standards (e.g. Darwin Core, ISO 19115, CSDGM, EML) are suitable for such specific purposes as WISER and suggest at least the linkage with standard metadata fields. Furthermore, we discuss whether the simple metadata documentation is enough for others to reuse a dataset and why there is still reluctance to publish both metadata and primary research data (i.e. time and financial constraints, misuse of data, abandoning intellectual property rights). We emphasise that metadata publication has major advantages as it makes datasets detectable by other scientists and generally makes a scientist’s work more visible.

Sensitivity of Spring Phenology to Warming Across Temporal and Spatial Climate Gradients in Two Independent Databases

Disparate ecological datasets are often organized into databases post hoc and then analyzed and interpreted in ways that may diverge from the purposes of the original data collections. Few studies, however, have attempted to quantify how biases inherent in these data (for example, species richness, replication, climate) affect their suitability for addressing broad scientific questions, especially in under-represented systems (for example, deserts, tropical forests) and wild communities. Here, we quantitatively compare the sensitivity of species first flowering and leafing dates to spring warmth in two phenological databases from the Northern Hemisphere. One—PEP725—has high replication within and across sites, but has low species diversity and spans a limited climate gradient. The other—NECTAR—includes many more species and a wider range of climates, but has fewer sites and low replication of species across sites. PEP725, despite low species diversity and relatively low seasonality, accurately captures the magnitude and seasonality of warming responses at climatically similar NECTAR sites, with most species showing earlier phenological events in response to warming. In NECTAR, the prevalence of temperature responders significantly declines with increasing mean annual temperature, a pattern that cannot be detected across the limited climate gradient spanned by the PEP725 flowering and leafing data. Our results showcase broad areas of agreement between the two databases, despite significant differences in species richness and geographic coverage, while also noting areas where including data across broader climate gradients may provide added value. Such comparisons help to identify gaps in our observations and knowledge base that can be addressed by ongoing monitoring and research efforts. Resolving these issues will be critical for improving predictions in understudied and under-sampled systems outside of the temperature seasonal mid-latitudes.

Pan-European distribution modelling of stream riparian zones based on multi-source Earth Observation data

Freshwater ecosystems are among the most degraded and threatened globally. A need for systematic assessment of riverine habitats is thus well-documented. Riparian zones are especially important due to the large array of ecosystem and social services they can provide, while also recently gaining a major role in the new European biodiversity policy. In this context, the necessity to gather spatial information on extent, distribution and characteristics of the riparian zones is clear. This paper presents the development of a novel model to derive geographical distribution and basic characteristics of stream riparian zones in Europe, including both river-floodplain systems and the riparian networks of minor and ephemeral streams. A series of relevant descriptive attributes (water boundary, vegetation presence, land-cover, upland boundary and local geomorphology) is initially selected from the existing literature to describe the presence of riparian zones. Spatially explicit estimates of these attributes are obtained using available Earth Observation data and pan-European thematic datasets. Finally, the information layers are combined through an aggregation system that assigns a degree of belonging to the riparian zone class using two fuzzy membership functions to evaluate water influence and presence of natural vegetation, respectively. Riparian zones with no hydrological connection are also incorporated in by the model based on functional considerations. The distribution of stream riparian zones was derived and mapped for the whole of Europe at fine resolution. Modelled riparian zones extend for approximately 91,000km2, considering the entire range of low to high membership to the riparian class. A characterization of land-cover types was derived based on Corine Land Cover 2000 data, showing that European riparian zones are strongly dominated by forest habitats. Accuracy assessment was performed using independent ecological datasets and visual validation, indicating that producer accuracy is equal to 84.5±1.3% and user accuracy to 72.6±5.8% at 95% confidence level. The proposed model and output can represent valuable information for large-scale research activities of riparian environments and to support national and supra-national conservation programmes.

Experimental investigation of false positive errors in auditory species occurrence surveys

False positive errors are a significant component of many ecological data sets, which in combination with false negative errors, can lead to severe biases in conclusions about ecological systems. We present results of a field experiment where observers recorded observations for known combinations of electronically broadcast calling anurans under conditions mimicking field surveys to determine species occurrence. Our objectives were to characterize false positive error probabilities for auditory methods based on a large number of observers, to determine if targeted instruction could be used to reduce false positive error rates, and to establish useful predictors of among-observer and among-species differences in error rates. We recruited 31 observers, ranging in abilities from novice to expert, who recorded detections for 12 species during 180 calling trials (66,960 total observations). All observers made multiple false positive errors, and on average 8.1% of recorded detections in the experiment were false positive errors. Additional instruction had only minor effects on error rates. After instruction, false positive error probabilities decreased by 16% for treatment individuals compared to controls with broad confidence interval overlap of 0 (95% CI:--46 to 30%). This coincided with an increase in false negative errors due to the treatment (26%;--3 to 61%). Differences among observers in false positive and in false negative error rates were best predicted by scores from an online test and a self-assessment of observer ability completed prior to the field experiment. In contrast, years of experience conducting call surveys was a weak predictor of error rates. False positive errors were also more common for species that were played more frequently but were not related to the dominant spectral frequency of the call. Our results corroborate other work that demonstrates false positives are a significant component of species occurrence data collected by auditory methods. Instructing observers to only report detections they are completely certain are correct is not sufficient to eliminate errors. As a result, analytical methods that account for false positive errors will be needed, and independent testing of observer ability is a useful predictor for among-observer variation in observation error rates.

Collinearity: a review of methods to deal with it and a simulation study evaluating their performance

Collinearity refers to the non independence of predictor variables, usually in a regression‐type analysis. It is a common feature of any descriptive ecological data set and can be a problem for parameter estimation because it inflates the variance of regression parameters and hence potentially leads to the wrong identification of relevant predictors in a statistical model. Collinearity is a severe problem when a model is trained on data from one region or time, and predicted to another with a different or unknown structure of collinearity. To demonstrate the reach of the problem of collinearity in ecology, we show how relationships among predictors differ between biomes, change over spatial scales and through time. Across disciplines, different approaches to addressing collinearity problems have been developed, ranging from clustering of predictors, threshold‐based pre‐selection, through latent variable methods, to shrinkage and regularisation. Using simulated data with five predictor‐response relationships of increasing complexity and eight levels of collinearity we compared ways to address collinearity with standard multiple regression and machine‐learning approaches. We assessed the performance of each approach by testing its impact on prediction to new data. In the extreme, we tested whether the methods were able to identify the true underlying relationship in a training dataset with strong collinearity by evaluating its performance on a test dataset without any collinearity. We found that methods specifically designed for collinearity, such as latent variable methods and tree based models, did not outperform the traditional GLM and threshold‐based pre‐selection. Our results highlight the value of GLM in combination with penalised methods (particularly ridge) and threshold‐based pre‐selection when omitted variables are considered in the final interpretation. However, all approaches tested yielded degraded predictions under change in collinearity structure and the ‘folk lore’‐thresholds of correlation coefficients between predictor variables of |r| >0.7 was an appropriate indicator for when collinearity begins to severely distort model estimation and subsequent prediction. The use of ecological understanding of the system in pre‐analysis variable selection and the choice of the least sensitive statistical approaches reduce the problems of collinearity, but cannot ultimately solve them.

Advances in global change research require open science by individual researchers

AbstractUnderstanding how species and ecosystems respond to climate change requires spatially and temporally rich data for a diverse set of species and habitats, combined with models that test and predict responses. Yet current study is hampered by the long‐known problems of inadequate management of data and insufficient description of analytical procedures, especially in the field of ecology. Despite recent institutional incentives to share data and new data archiving infrastructure, many ecologists do not archive and publish their data and code. Given current rapid rates of global change, the consequences of this are extreme: because an ecological dataset collected at a certain place and time represents an irreproducible set of observations, ecologists doing local, independent research possess, in their file cabinets and spreadsheets, a wealth of information about the natural world and how it is changing. Although large‐scale initiatives will increasingly enable and reward open science, we believe that change demands action and personal commitment by individuals – from students and PIs. Herein, we outline the major benefits of sharing data and analytical procedures in the context of global change ecology, and provide guidelines for overcoming common obstacles and concerns. If individual scientists and laboratories can embrace a culture of archiving and sharing we can accelerate the pace of the scientific method and redefine how local science can most robustly scale up to globally relevant questions.

Using model‐data fusion to interpret past trends, and quantify uncertainties in future projections, of terrestrial ecosystem carbon cycling

AbstractUncertainties in model projections of carbon cycling in terrestrial ecosystems stem from inaccurate parameterization of incorporated processes (endogenous uncertainties) and processes or drivers that are not accounted for by the model (exogenous uncertainties). Here, we assess endogenous and exogenous uncertainties using a model‐data fusion framework benchmarked with an artificial neural network (ANN). We used 18 years of eddy‐covariance carbon flux data from theHarvard forest, where ecosystem carbon uptake has doubled over the measurement period, along with 15 ancillary ecological data sets relative to the carbon cycle. We test the ability of combinations of diverse data to constrain projections of a process‐based carbon cycle model, both against the measured decadal trend and under future long‐term climate change. The use of high‐frequency eddy‐covariance data alone is shown to be insufficient to constrain model projections at the annual or longer time step. Future projections of carbon cycling under climate change in particular are shown to be highly dependent on the data used to constrain the model. Endogenous uncertainties in long‐term model projections of future carbon stocks and fluxes were greatly reduced by the use of aggregated flux budgets in conjunction with ancillary data sets. The data‐informed model, however, poorly reproduced interannual variability in net ecosystem carbon exchange and biomass increments and did not reproduce the long‐term trend. Furthermore, we use the model‐data fusion framework, and theANN, to show that the long‐term doubling of the rate of carbon uptake atHarvard forest cannot be explained by meteorological drivers, and is driven by changes during the growing season. By integrating all available data with the model‐data fusion framework, we show that the observed trend can only be reproduced with temporal changes in model parameters. Together, the results show that exogenous uncertainty dominates uncertainty in future projections from a data‐informed process‐based model.

Regression with Empirical Variable Selection: Description of a New Method and Application to Ecological Datasets

Despite recent papers on problems associated with full-model and stepwise regression, their use is still common throughout ecological and environmental disciplines. Alternative approaches, including generating multiple models and comparing them post-hoc using techniques such as Akaike's Information Criterion (AIC), are becoming more popular. However, these are problematic when there are numerous independent variables and interpretation is often difficult when competing models contain many different variables and combinations of variables. Here, we detail a new approach, REVS (Regression with Empirical Variable Selection), which uses all-subsets regression to quantify empirical support for every independent variable. A series of models is created; the first containing the variable with most empirical support, the second containing the first variable and the next most-supported, and so on. The comparatively small number of resultant models (n = the number of predictor variables) means that post-hoc comparison is comparatively quick and easy. When tested on a real dataset – habitat and offspring quality in the great tit (Parus major) – the optimal REVS model explained more variance (higher R2), was more parsimonious (lower AIC), and had greater significance (lower P values), than full, stepwise or all-subsets models; it also had higher predictive accuracy based on split-sample validation. Testing REVS on ten further datasets suggested that this is typical, with R2 values being higher than full or stepwise models (mean improvement = 31% and 7%, respectively). Results are ecologically intuitive as even when there are several competing models, they share a set of “core” variables and differ only in presence/absence of one or two additional variables. We conclude that REVS is useful for analysing complex datasets, including those in ecology and environmental disciplines.