Guest blog post by Jürgen Dengler, Florian Jansen & François Gillet
It took seven years and hundreds of hours of work by an international team of 34 authors to develop and publish the most comprehensive system of ecological indicator values (EIVs) of vascular plants in Europe to date.
EIVE 1.0 is now available as an open access database and described in the accompanying paper (Dengler et al. 2023).
EIVE 1.0 provides the five most-used ecological indicators, M – moisture, N – nitrogen, R – reaction, L – light and T – temperature, for a total of 14,835 vascular plant taxa in Europe, or between 13,748 and 14,714 for the individual indicators. For each of these taxa, EIVE contains three values: the EIVE niche position indicator, the EIVE niche width indicator and the number of regional EIV systems on which the assessment was based. Both niche position and niche width are given on a continuous scale from 0 to 10, not as categorical ordinal values as in the source systems.
Evidently, EIVE can be an important tool for continental-scale analyses of vegetation and floristic data in Europe.
It will allow to analyse the nearly 2 million vegetation plots currently contained in the European Vegetation Archive (EVA; Chytrý et al. 2016) in new ways.
Since EVA apart from elevation, slope inclination and aspect hardly contains any in situ measured environmental variables, the numerous macroecological studies up to date had to rely on coarse modelled environmental data (e.g. climate) instead. This is particularly problematic for soil variables such as pH, moisture or nutrients, which can change dramatically within a few metres.
Here, the approximation of site conditions by mean ecological indicator values can improve the predictive power substantially (Scherrer and Guisan 2019). Likewise, in broad-scale vegetation classification studies, mean EIVE values per plot would allow a better characterisation of the distinguished vegetation units. Lastly, one should not forget that most countries in Europe do not have a national EIV system, and here EIVE could fill the gap.
Almost on the same day as EIVE 1.0 another supranational system of ecological indicator values in Europe has been published by Tichý et al. (2023) with a similar approach.
Thus, it will be important for vegetation scientists in Europe to understand the pros and cons of both systems to allow the wise selection of the most appropriate tool:
- EIVE 1.0 is based on 31 regional EIV systems, while Tichý et al. (2023) uses 12.
- Both systems provide indicator values for moisture, nitrogen/nutrients, reaction, light and temperature, while Tichý et al. (2023) additionally has a salinity indicator.
- Tichý et al. (2023) aimed at using the same scales as Ellenberg et al. (1991), which means that the scales vary between indicators (1–9, 0–9, 1–12), while EIVE has a uniform interval scale of 0–10 for all indicators.
- Only EIVE provides niche width in addition to niche position. Niche width is an important aspect of the niche and might be used to improve the calculation of mean indicator values per plot (e.g. by weighting with inverse niche width).
- The taxonomic coverage is larger in EIVE than in Tichý et al. (2023): 14,835 vs. 8,908 accepted taxa and 11,148 vs. 8,679 species.
- EIVE provides indicator values for accepted subspecies, while Tichý et al. (2023) is restricted to species and aggregates. Separate indicator values for subspecies might be important for two reasons: (a) subspecies often strongly differ in at least one niche dimension; (b) many of the taxa now considered as subspecies have been treated at species level in the regional EIV systems.
- Tichý et al. (2023) added 431 species not contained in any of the source systems based on vegetation-plot data from the European Vegetation Archive (EVA; Chytrý et al. 2016) while EIVE calculated the European indicator values only for taxa occurring at least in one source system.
- While both systems present maps that suggest a good coverage across Europe, Tichý et al. (2023)’s source systems largely were from Central Europe, NW Europe and Italy, but, unlike EIVE, these authors did not use source systems from the more “distal” parts of Europe, such as Sweden, Faroe Islands, Russia, Georgia, Romania, Poland and Spain, and they used only a small subset of indicators of the EIV systems of Ukraine, Greece and the Alps.
- In a validation with GBIF-derived data on temperature niches, Dengler et al. (2023) showed that EIVE has a slightly stronger correlation than Tichý et al. (2023)’s indicators (r = 0.886 vs. 0.852).
How did EIVE manage to integrate all EIV systems in Europe that contained at least one of the selected indicators for vascular plants, while Tichý et al. (2023) used only a small subset?
This difference is mainly due to a more complex workflow in EIVE (which also was one of the reasons why the preparation took so long). First, Tichý et al. (2023) restricted their search to EIV systems and indicators that had the same number of categories as the “original” Ellenberg system.
Second, from these they discarded those that showed a too low correlation with Ellenberg. By contrast, EIVE’s workflow allowed the use of any system with an ordinal (or even metric) scale, irrespective of the number of categories or the initial match with Ellenberg et al. (1991).
EIVE also did not treat one system (Ellenberg) as the master to assess all others but considered each of them equally valid. While indeed the individual EIV systems are often quite inconsistent, i.e. even if they refer to Ellenberg, the same value of an indicator in one system might mean something different in another system, our iterative linear optimisation enabled us to adjust all 31 systems for the five indicators to a common basis.
This in turn allowed deriving EIVE as the consensus system of all the source systems. The fact that in our validation of the temperature indicator, EIVE performed better than Tichý et al. (2023) and much better than most of the regional EIV systems might be attributable to the so-called “wisdom of the crowd”, going back to the statistician Francis Galton who found that averaging numerous independent assessments (even by laymen) of a continuous quantity can leads to very good estimates of the true value.
Apart from the indicator values themselves, EIVE has a second main feature that might not be so obvious at first glance, but which actually took the EIVE team, including several taxonomists, more time than the workflow to generate the indicator values themselves: the taxonomic backbone. EIVE for vascular plants is fully based on the taxonomic concept (including the synonymic relationships) of the Euro+Med Plantbase.
However, since Euro+Med lacks an important part of taxa that are frequently recorded in vegetation plots, to make our backbone fully usable to vegetation science, we expanded it beyond Euro+Med to something called “Euro+Med augmented”. We particularly added hybrids, neophytes and aggregates, three groups of plants hitherto only very marginally covered in Euro+Med. All additions were done by experts consistently with the taxonomic concept of Euro+Med and are fully documented. Likewise, many additional synonym relationships had to be added that were missing in Euro+Med.
Finally, we implemented the so-called “concept synonymy” (see Jansen and Dengler 2010), which allows the assignment of the same name from different sources to different accepted names (“taxonomic concepts”). This applies mainly to nested taxa that are treated at different levels in different sources, e.g. once as species with several subspecies, once as aggregate with several species. However, there are also some cases of misapplied names (i.e. names that were not used in agreement with their nomenclatural type in certain EIV systems). Such cases generally cannot be solved by the various tools for automatic taxonomic cleaning, but require experts who make a case-by-case decision.
The whole taxonomic workflow of EIVE is fully transparent with an R code that “digests”:
(a) the names as they are in the source systems,
(b) the official Euro+Med database and
(c) tables that document our additions and modifications (with reasons and references).
This comprehensive documentation will allow continuous and efficient improvement in the future, be it because of taxonomic novelties adopted in Euro+Med or because EIVE’s experts decide to change certain interpretations. That way, “Euro+Med augmented” and the accompanying R-based workflow can also be a valuable tool for other projects that wish to harmonise plant taxonomic information from various sources at a continental scale, e.g. in vegetation-plot databases such as GrassPlot (Dengler et al. 2018) and EVA (Chytrý et al. 2016).
The publication of EIVE 1.0 is not the endpoint, but rather a starting point for future developments in a community-based approach.
Together with interested colleagues from outside, the EIVE core team plans to prepare better and more comprehensive releases of EIVE in the future, including updates to its taxonomic backbone.
Future releases of EIVE will be published in fixed versions, typically together with a paper that describes the changes in the content.
As steps for the next two years, we anticipate that we will first add further taxa (bryophytes, lichens, macroalgae) and some additional indicators, both of which are relatively easy with our established R-based workflow. Then we plan EIVE 2.0 that will use the approx. 2 million vegetation plots in EVA (Chytrý et al. 2016) to re-calibrate EIVE for all taxa (see http://euroveg.org/requests/EVA-data-request-form-2022-02-10-Dengleretal.pdf).
This Behind the paper post refers to the article Ecological Indicator Values for Europe (EIVE) 1.0 by Jürgen Dengler, Florian Jansen, Olha Chusova, Elisabeth Hüllbusch, Michael P. Nobis, Koenraad Van Meerbeek, Irena Axmanová, Hans Henrik Bruun, Milan Chytrý, Riccardo Guarino, Gerhard Karrer, Karlien Moeys, Thomas Raus, Manuel J. Steinbauer, Lubomir Tichý, Torbjörn Tyler, Ketevan Batsatsashvili, Claudia Bita-Nicolae, Yakiv Didukh, Martin Diekmann, Thorsten Englisch, Eduardo Fernandez Pascual, Dieter Frank, Ulrich Graf, Michal Hájek, Sven D. Jelaska, Borja Jiménez-Alfaro, Philippe Julve, George Nakhutsrishvili, Wim A. Ozinga, Eszter-Karolina Ruprecht, Urban Šilc, Jean-Paul Theurillat, and François Gillet published in Vegetation Classification and Survey (https://doi.org/10.3897/VCS.98324).
Follow the Vegetation Classification and Survey journal on Facebook and Twitter.
Brief personal summaries:
Jürgen Dengler is a Professor of Vegetation Ecology at the Zurich University of Applied Science (ZHAW) in Wädenswil, Switzerland. Among others, he cofounded the European Vegetation Database (EVA), the global vegetation-plot database “sPlot” and the “GrassPlot” database of the Eurasian Dry Grassland Group. His major research interests are grassland ecology, grassland conservation, biodiversity patterns, macroecology, vegetation change, broad-scale vegetation classification, methodological developments in vegetation ecology and ecoinformatics.
Florian Jansen is a Professor of Landscape Ecology at the University of Rostock, Germany. His research interests are vegetation ecology and dynamics, mire ecology including greenhouse gas emissions, and numerical ecology with R. He (co-)founded the German Vegetation Database vegetweb.de, the European Vegetation Database (EVA), and the global vegetation-plot database “sPlot”. He wrote the R package eHOF for modelling species response curves along one-dimensional ecological gradients.
François Gillet is an Emeritus Professor of Community Ecology at the University of Franche-Comté in Besançon, France. His major research interests are vegetation diversity, ecology and dynamics, grassland and forest ecology, integrated synusial phytosociology, numerical ecology with R, dynamic modelling of social-ecological systems.
Chytrý, M., Hennekens, S.M., Jiménez-Alfaro, B., Knollová, I., Dengler, J., Jansen, F., Landucci, F., Schaminée, J.H.J., Aćić, S., (…) & Yamalov, S. 2016. European Vegetation Archive (EVA): an integrated database of European vegetation plots. Applied Vegetation Science 19: 173–180.
Dengler J, Wagner V, Dembicz I, García-Mijangos I, Naqinezhad A, Boch S, Chiarucci A, Conradi T, Filibeck G, … Biurrun I (2018) GrassPlot – a database of multi-scale plant diversity in Palaearctic grasslands. Phytocoenologia 48: 331–347.
Dengler, J., Jansen, F., Chusova, O., Hüllbusch, E., Nobis, M.P., Van Meerbeek, K., Axmanová, I., Bruun, H.H., Chytrý, M., (…) & Gillet, F. 2023. Ecological Indicator Values for Europe (EIVE) 1.0. Vegetation Classification and Survey 4: 7–29.
Ellenberg H, Weber HE, Düll R, Wirth V, Werner W, Paulißen D (1991) Zeigerwerte von Pflanzen in Mitteleuropa. Scripta Geobotanica 18: 1–248.
Jansen F, Dengler J (2010) Plant names in vegetation databases – a neglected source of bias. Journal of Vegetation Science 21: 1179–1186.
Midolo, G., Herben, T., Axmanová, I., Marcenò, C., Pätsch, R., Bruelheide, H., Karger, D.N., Acic, S., Bergamini, A., Bergmeier, E., Biurrun, I., Bonari, G., Carni, A., Chiarucci. A., De Sanctis, M., Demina, O., (…), Dengler, J., (…) & Chytrý, M. 2023. Disturbance indicator values for European plants. Global Ecology and Biogeography 32: 24–34.
Scherrer D, Guisan A (2019) Ecological indicator values reveal missing predictors of species distributions. Scientific Reports 9: Article 3061.
Tichý, L, Axmanová, I., Dengler, J., Guarino, R., Jansen, F., Midolo, G., Nobis, M.P., Van Meerbeek, K., Aćić, S., (…) & Chytrý, M. 2023. Ellenberg-type indicator values for European vascular plant species. Journal of Vegetation Science 34: e13168.