This Special Feature includes contributions on data-processing of large ecological datasets under the heading ecoinformatics. Herewith the latter term is now also established in the Journal of Vegetation Science. Ecoinfomatics is introduced as a rapid growing field within community ecology which is generating exciting new developments in ecology and in particular vegetation ecology. In our field, ecoinformatics deals with the understanding of patterns of species distributions at local and regional scales, and on the assemblages of species in relation to their properties, the local environment and their distribution in the region. Community ecology using ecoinformatics is related to bioinformatics, community ecology, biogeography and macroecology.

We make clear how ecoinformatics in vegetation science and particularly the IAVS Working Group on Ecoinformatics has developed from the work of the old Working Group for Data Processing which was active during the 1970s and 1980s. Recent developments, including the creation of TURBOVEG and SynBioSys in Europa and VEGBANK in the USA, form a direct link with these pioneer activities, both scientifically and personally.

The contributions collected in this Special Feature present examples of seco-infeveral types of the use of databases and the application of programmes and models. The main types are the study of long-term vegetation dynamics in different cases of primary and secondary succession and the understanding of successional developments in terms of species traits.

Among the future developments of great significance we mention the use of a variety of different large datasets for the study of the distribution and ecology and conservation of rare and threatened species.

The rapid developments in computer techniques and the availability of large datasets open new perspectives for vegetation analysis aiming at better understanding of the ecology and functioning of ecosystems and underlying mechanisms. Information systems prove to be helpful tools in this new field. Such information systems may integrate different biological levels, viz. species, community and landscape. They incorporate a GIS platform for the visualization of the various layers of information, enabling the analysis of patterns and processes which relate the individual levels. An example of a newly developed information system is SynBioSys Europe, an initiative of the European Vegetation Survey (EVS). For the individual levels of the system, specific sources are available, notably national and regional Turboveg databases for the community level and data from the recently published European Map of Natural Vegetation for the landscape level. The structure of the system and its underlying databases allow user-defined queries. With regard to its application, such information systems may play a vital role in European nature planning, such as the implementation the EU-program Natura 2000. To illustrate the scope and perspectives of the program, some examples from The Netherlands are presented. They are dealing with long-term changes in grassland ecosystems, including shifts in distribution, floristic composition, and ecological indicator values.

Questions: Is change in cover of dominant species driving the velocity of succession or is it species turnover (1)? Is the length of the time-step chosen in sampling affecting our recognition of the long-term rate of change (2)?

Location: 74 permanent plots located in the Swiss National Park, SE Switzerland, ca. 1900 m a.s.l.

Methods: We superimpose several time-series from permanent plots to one single series solely based on compositional dissimilarity. As shown earlier (Wildi & Schütz 2000) this results in a synthetic series covering about 400 to 650 yr length. Continuous power transformation of cover-percentage scores is used to test if the dominance or the presence-absence of species is governing secondary succession from pasture to forest. The effect of time step length is tested by sub-samples of the time series.

Results: Altering the weight of presence-absence versus dominance of species affects the emerging time frame, while altering time step length is uncritical. Where species turnover is fast, different performance scales yield similar results. When cover change in dominant species prevails, the solutions vary considerably. Ordinations reveal that the synthetic time series seek for shortest paths of the temporal pattern whereas in the real system longer lasting alternatives exist.

Conclusions: Superimposing time series differs from the classical space-for-time substitution approach. It is a computation-based method to investigate temporal patterns of hundreds of years fitting between direct monitoring (usually limited to decades) and the analysis of proxy-data (for time spans of thousands of years and more).

Nomenclature: Hess et al. (1976)

Questions: Primary succession, measured by changes in species composition, is slow, usually forcing a chronosequence approach. A unique data set is used to explore spatial and temporal changes in vegetation structure after a 1980 volcanic eruption. On the basis of data from a transect of 20 permanent plots with an altitudinal range of 250 m sampled through 2005, two questions are asked: Do changes along the transect recapitulate succession? Do plots converge to similar composition over time?

Location: A ridge between 1218 and 1468 m on Mount St. Helens, Washington, USA.

Methods: Repeat sampling of plots for species cover along a 1-km transect. Floristic changes were characterized by techniques including DCA, clustering and similarity.

Results: Species richness and cover increased with time at rates that decreased with increasing elevation. The establishment of Lupinus lepidus accelerated the rate of succession and may control its trajectory. Diversity (H′) at first increased with richness, then declined as dominance hierarchies developed. Primary succession was characterized by overlapping phases of species assembly (richness), vegetation maturation (diversity peaks, cover expands) and inhibition (diversity declines). Each plot passed through several community classes, but by 2005, only four classes persisted. Succession trajectories (measured by DCA) became shorter with elevation. Similarity between groups of plots defined by their classification in 2005 did not increase with time. Similarity within plot groups converged slightly at the lower elevations. Despite similarities between temporal and spatial trends in composition, trajectories of higher plots do not recapitulate those of lower plots, apparently because Lupinus was not an early colonist. Any vegetation convergence has been limited to plots that are in close proximity.

Nomenclature: A list of species used in this study is at  http://protist.biology.washington.edu/-delmoral/. Nomenclature follows the Integrated Taxonomic Information System:  http://www.itis.usda.gov/advanced_search.html.

Questions: 1. Which plant traits and habitat characteristics best explain local above-ground persistence of vascular plant species and 2. Is there a trade-off between local above-ground persistence and the ability for seed dispersal and below-ground persistence in the soil seed bank?

Locations: 845 long-term permanent plots in terrestrial habitats across the Netherlands.

Methods: We analysed the local above-ground persistence of vascular plants in permanent plots (monitored once a year for ca. 16 year) with respect to functional traits and habitat preferences using survival statistics (Kaplan-Meier analysis and Cox' regression). These methods account for censored data and are rarely used in vegetation ecology.

Results: Local above-ground persistence is determined by both functional traits (especially the ability to form long-lived clonal connections) and habitat preferences (especially nutrient requirements). Above-ground persistence is negatively related to the ability for dispersal by wind and to the ability to accumulate a long-term persistent soil seed bank (‘dispersal through time’) and is positively related to the ability for dispersal by water.

Conclusions: Most species have a half-life expectation over 15 years, which may contribute to time lags after changes in habitat quality or -configuration (‘extinction debt’). There is evidence for a trade-off relationship between local above-ground persistence and below-ground seed persistence, while the relationship with dispersal in space is vector specific. The rate of species turnover increases with productivity.

Nomenclature: van der Meijden (1990).

Questions: 1. Do relationships among forest plant traits correspond to dispersability-persistence trade-offs or other inter-trait correlations found in the literature? 2. Do species groups delineated by trait similarity, differ in occurrence in ancient vs. new forests or isolated vs more continuous forest patches? 3. Are these patterns consistent for different forest types?

Location: Central Belgium, near Leuven.

Methods: We investigate the distributions of a large set of plant traits and combinations among all forest species occurring in patches with varying forest continuity and isolation. Through calculation of Gower's similarity index and subsequent clustering, ‘emergent’ species groups are delineated. Then, the relative occurrence of these different groups in forest patches of different age and size, sustaining different forest types (alluvial vs. Quercion), and having different isolation status is compared through multivariate GLM analysis.

Results: Correlations among several life history traits point towards trade-offs of dispersability and fecundity vs. longevity. We distinguished three species groups: 1= mainly shrubs or climbers with fleshy or wind dispersed fruits and high dispersal potential; 2 = dominated by small, mainly vegetatively reproducing herbs; 3 = with spring flowering herbs with large seeds and mainly unassisted dispersal. Relative occurrence of these groups was significantly affected by forest age, area, isolation and forest type. Separate analyses for alluvial and Quercion forests indicated that the relative importance of these factors may differ, depending on forest type and species group. Both forest continuity and isolation are important in restricting the relative occurrence of forest species in alluvial forests, whatever their group membership. In Quercion forests forest patch area was the primary determinant of relative occurrence of species groups.

Conclusions: It is very important to preserve the actual forest area including the spatial setting and the dispersal infrastructure within the landscape. Next, forest connectivity may be restored, but it is inherently a long process.

Nomenclature: Lambinon et al. (1998).

Problem: A series of long-term field experiments is described, with particular reference to monitoring and quality control. This paper addresses problems in data-management of particular importance for long-term studies, including data manipulation, archiving, quality assessment, and flexible retrieval for analysis

Method: The problems were addressed using a purpose-built database system, using commercial software and running under Microsoft Windows.

Conclusion: The database system brings many advantages compared to available software, including significantly improved quality checking and access. The query system allows for easy access to data sets thus improving the efficiency of analysis. Quality assessments of the initial dataset demonstrated that the database system can also provide general insight into types and magnitudes of error in data-sets. Finally, the system can be generalised to include data from a number of different projects, thus simplifying data manipulation for meta-analysis.

Nomenclature: Stace (1997).

Question: Can the distribution and abundance of Vaccinium myrtillus be reasonably predicted with soil nutritional and climatic factors?

Location: Forests of France.

Methods: We used Braun-Blanquet abundance/dominance information for Vaccinium myrtillus on 2905 forest sites extracted from the phyto-ecological database EcoPlant, to characterize the species ecological response to climatic and edaphic factors and to predict its cover/abundance at the national scale. The link between cover/abundance of the species and climatic (65 monthly and annual predictors concerning temperature, precipitation, radiation, potential evapotranspiration, water balance) and edaphic (two predictors: soil pH and C:N ratio) factors was investigated with proportional odds models. We evaluated the quality of our model with 9830 independent relevés extracted from Sophy, a large phytosociological database for France.

Results: In France, Vaccinium myrtillus is at the southern limit of its European geographic range and three environmental factors (mean annual temperature, soil pH and C:N ratio) allow prediction of its distribution and abundance in forests with high success rates. The species reveals a preference for colder sites (especially mountains) and nutritionally poor soils (low pH and high C:N ratio). A predictive map of its geographic range reveals that the main potential habitats are mountains and northwestern France. The potential habitats with maximal expected abundance are the Vosges and the Massif central mountains, which are both acidic mountains.

Conclusions: Complete niche models including climate and soil nutritional conditions allow an improvement of the spatial prediction of plant species abundance at a broad scale. The use of soil nutritional variables in distribution models further leads to an improvement in the prediction of plant species habitats within their geographical range.

Question: How well can mortality probabilities of deciduous trees (Fagus sylvatica) and conifers (Abies alba) be predicted using permanent plot data that describe growth patterns, tree species, tree size and site conditions?

Location: Fagus forests in the montane belt of the Jura folds (Switzerland).

Method: Permanent plot data were used to develop and validate logistic regression models predicting survival probabilities of individual trees. Backward model selection led to a reduced model containing the growth-related variable ‘relative basal area increment’ (growth-dependent mortality) and variables not directly reflecting growth such as species, size and site (growth-independent mortality).

Results: The growth-mortality relationship was the same for both species (growth-dependent mortality). However, species, site and tree size also influenced mortality probabilities (growth-independent mortality). The predicted survival probabilities of the final model were well calibrated, and the model showed an excellent discriminatory power (area under the receiver operating characteristic curve = 0.896).

Conclusion: Mortality probabilities of Fagus sylvatica and Abies alba can be predicted with high discriminatory power using a well calibrated logistic regression model. Extending this case study to a larger number of tree species and sites could provide species- and site-specific tree mortality models that allow for more realistic projections of forest succession.

Problem: Data from over 100 permanent sample plots which have been studied for 10–20 years need a suitable system for storage which allows simple data manipulation and retrieval for analysis.

Methods: A relational database linking tree records, taxonomic nomenclature and corresponding environmental data has been built in MS Access as part of the RAINFOR project.

Conclusion: The database allows flexible and long-term use of a large amount of data: more than 100 tree plots across Amazonia, incorporating over 80 000 records of individual trees and over 300 000 total records of tree diameter measurements from successive censuses. The database is designed to enable linkages to existing soil, floristic or plant-trait databases. This database will be a useful tool for exploring the impact of environmental factors on forest structure and dynamics at local to continental scales, and long term changes in forest ecology. As an early example of its potential, we explore the impact of different methodological assumptions on estimates of tropical forest biomass and carbon storage.

Questions: Are community dynamics in old-growth forests predictable? Convergent? Equilibrial? Are answers to these questions dependent on temporal and spatial scale? How can complex, long-term observational data be used most powerfully to address these questions?

Location: 100-ha tract of old-growth cool-temperate forest in northern Michigan, USA.

Methods: Woody stems were measured, on 243 permanent plots, several times, at varying intervals and intensity, over 70 years. A range of visualization tools and multivariate statistics were used to extract patterns and address questions posed.

Results: This ancient forest is not equilibrial; compositional trends suggest that changes are competitively driven and reflect long-lasting effects of disturbance. Predictability of community change varies across environmental gradients, with interval between samples, with spatial scale, and depending on type of predictability being assessed. Plot trajectories in species-space and changes in diversity suggest successional convergence within some habitats, but not across habitats. Dynamics are strongly structured at the scale of ‘habitat-patches’.

Conclusions: Appropriate address of questions about com­munity dynamics requires observational data of appropriate spatial and temporal scale and resolution. Powerful use of such data-sets calls for data-management and analysis tools that are robust with respect to irregularities in design and data structure. While interpretation of long-term descriptive data is challenging, appropriate analyses cast light on late successional dynamics, allowing address of models and hypotheses that are otherwise difficult to test.

Nomenclature: Gleason & Cronquist (1991).

Questions: 1. Is the above-ground biomass in natural temperate forests positively correlated with tree species diversity?

2. Is this biomass related to the diversity of tree functional groups?

Location: We used published data from over 100 permanent plots located in natural temperate forests in the Czech Republic, Poland and Slovakia.

Methods: We related the number of tree species and Simpson's index of tree species diversity to the above-ground biomass in natural forest stands, and we repeated the same calculations for the identification of functional groups of trees using PCA analysis of functional traits.

Results: Analysed sites ranged from almost pure subalpine spruce stands to mixed deciduous lowland forests with eight tree species per stand. The above-ground biomass accumulation ranged from 169 to 536 tons of dry mass per hectare. For the analysed data set the relationship between tree species diversity and biomass accumulation was not significant but showed a negative trend. Similar results were obtained in analyses employing tree functional groups instead of tree species. A significant negative relationship was found after four stands located in the highest elevations had been removed from the data set.

Conclusions: There is a weak negative relationship between tree species diversity and above-ground biomass in natural forests of Central Europe.

Nomenclature: Tutin et al. (1964–1980).

Question: Is there any generality in terms of leaf trait correlations and the multiple role of leaf traits (response to and/or effect on) during secondary succession?

Location: A secondary successional sere was sampled at four different ages since abandonment from several years to nearly 150 years on the Loess Plateau of northwestern China.

Method: Specific leaf area (SLA), leaf mass per area (LMA), leaf nitrogen (N_mass, N_area), leaf phosphorus (P_mass, P_area) and leaf dry matter content (LDMC) were measured for all species recorded in the successional sere. Above-ground net primary productivity (ANPP) and specific rate of litter mass loss (SRLML) were measured as surrogates for ecosystem properties. Soil total carbon (C) and nitrogen (N) were measured in each stage. Leaf traits were related to ecosystem properties and soil nutrient gradients, respectively.

Results: LMA is correlated with N_area and P_area, and negatively with N_mass. Correlation between N_area and P_area was higher than between N_mass and P_mass. At the community level, field age, community hierarchy and their interaction explain 64.4 - 93.5% of the variation in leaf traits. At the species level, field age explains 22.4 - 45.5% of the variation in leaf traits (excl. P_area) while plant functional group has a significant effect only for N_mass. LDMC is correlated with ANPP and negatively with SRLML; P_mass is correlated with SRLML.

Conclusions: Mean values of LMA, N_mass and N_area are close to the worldwide means, suggesting that large-scale climate has a profound effect on leaf mass and leaf nitrogen allocation, while environmental gradients represented by succession have little influence on leaf-trait values. Correlations between leaf traits, such as LMA- N_area , LMA- P_area and LMA- N_mass shown in previous studies, are confirmed here. Although none of the leaf traits is proved to be both a response trait and an effect trait independent of time scale and community hierarchy, mass-based leaf N is likely a sensitive response trait to soil C and N gradients. In addition, LDMC can be a marker for ANPP and SRLML, while mass-based leaf P can be a marker for SRLML.

Nomenclature: Anon. (1959–2004).

Questions: Did the forest area in the Swiss Alps increase between 1985 and 1997? Does the forest expansion near the tree line represent an invasion into abandoned grasslands (ingrowth) or a true upward shift of the local tree line? What land cover / land use classes did primarily regenerate to forest, and what forest structural types did primarily regenerate? And, what are possible drivers of forest regeneration in the tree line ecotone, climate and/or land use change?

Location: Swiss Alps.

Methods: Forest expansion was quantified using data from the repeated Swiss land use statistics GEOSTAT. A moving window algorithm was developed to distinguish between forest ingrowth and upward shift. To test a possible climate change influence, the resulting upward shifts were compared to a potential regional tree line.

Results: A significant increase of forest cover was found between 1650 m and 2450 m. Above 1650 m, 10% of the new forest areas were identified as true upward shifts whereas 90% represented ingrowth, and we identified both land use and climate change as likely drivers. Most upward shift activities were found to occur within a band of 300 m below the potential regional tree line, indicating land use as the most likely driver. Only 4% of the upward shifts were identified to rise above the potential regional tree line, thus indicating climate change.

Conclusions: Land abandonment was the most dominant driver for the establishment of new forest areas, even at the tree line ecotone. However, a small fraction of upwards shift can be attributed to the recent climate warming, a fraction that is likely to increase further if climate continues to warm, and with a longer time-span between warming and measurement of forest cover.

Nomenclature: Aeschimann & Heitz (1996).

Question: Can satellite time series be used to identify tree and grass green-up dates in a semi-arid savanna system, and are there predictable environmental cues for green-up for each life form?

Location: Acacia nigrescens/Combretum apiculatum savanna, Kruger National Park, South Africa (25° S, 31° E).

Methods: Remotely-sensed data from the MODIS sensor were used to provide a five year record of greenness (NDVI) between 2000 and 2005. The seasonal and inter-annual patterns of leaf display of trees and grasses were described, using additional ecological information to separate the greening signal of each life form from the satellite time series. Linking this data to daily meteorological and soil moisture data allowed the cues responsible for leaf flush in trees and grasses to be identified and a predictive model of savanna leaf-out was developed. This was tested on a 22-year NDVI dataset from the Advanced Very High Resolution Radiometer.

A day length cue for tree green-up predicted 86% of the green-ups with an accuracy better than one month. A soil moisture and day length cue for grass green-up predicted 73% of the green-ups with an accuracy better than a month, and 82% within 45 days. This accuracy could be improved if the temporal resolution of the satellite data was shortened from the current two weeks.

Conclusions: The data show that at a landscape scale savanna trees have a less variable phenological cycle (within and between years) than grasses. Realistic biophysical models of savanna systems need to take this into account. Using climatic data to predict these dynamics is a feasible approach.

Question: Is ombrotrophic bog vegetation in an oceanic region of southwestern Sweden changing in the same direction over a five year period (1999– 2004) as northwest European bogs in the last 50 years, i.e. towards drier and more eutrophic vegetation?

Location: The province of Halland, southwestern Sweden.

Methods: Changes in species composition were monitored in 750 permanently marked plots in 25 ombrotrophic bogs from 1999 to 2004. Changes in species occurrences and richness were analysed and a multivariate statistical method (DCA) was used to analyse vegetation changes.

Results: The species composition changed towards wetter rather than drier conditions, which is unlike the general pattern of vegetation change on bogs in northwestern Europe. Species typical of wetter site conditions including most Sphagnum species increased in abundance on the bogs until 2004. The total number of species per plot increased, mostly due to the increased species richness of Sphagnum species. Nitrogen-demanding (eutrophic) species increased in occurrence.

Conclusions: Ombrotrophic bog vegetation in an oceanic region in Sweden became wetter and was resilient to short-term climatic shifts, after three years of below normal precipitation followed by several years with normal precipitation levels. Shifts towards more nitrogen demanding species were rapid in this region where the deposition levels have been high for several decades.

Nomenclature: Karlsson (1997) for vascular plants; Söderström & Hedenäs (1998) for bryophytes; Moberg et al. (1995) for lichens.

Question: Is it possible to mathematically classify relevés into vegetation types on the basis of their average indicator values, including the uncertainty of the classification?

Location: The Netherlands.

Method: A large relevé database was used to develop a method for predicting vegetation types based on indicator values. First, each relevé was classified into a phytosociological association on the basis of its species composition. Additionally, mean indicator values for moisture, nutrients and acidity were computed for each relevé. Thus, the position of each classified relevé was obtained in a three-dimensional space of indicator values. Fitting the data to so called Gaussian Mixture Models yielded densities of associations as a function of indicator values. Finally, these density functions were used to predict the Bayesian occurrence probabilities of associations for known indicator values. Validation of predictions was performed by using a randomly chosen half of the database for the calibration of densities and the other half for the validation of predicted associations.

Results and Conclusions: With indicator values, most relevés were classified correctly into vegetation types at the association level. This was shown using confusion matrices that relate (1) the number of relevés classified into associations based on species composition to (2) those based on indicator values. Misclassified relevés belonged to ecologically similar associations. The method seems very suitable for predictive vegetation models.