|
Getting your Trinity Audio player ready...
|
Post provided by Grzegorz Ostrowski, Severin Aicher, Agnieszka Mankiewicz & Jürgen Dengler, originally posted to vegsciblog.org.
Mean ecological indicator values (EIVs) are widely used by vegetation ecologists throughout Europe. They allow for an efficient assessment of site conditions (bioindication) of vegetation plots when measurements of the physical, chemical or land use conditions would be too costly or time-consuming or not possible at all, for example, for the millions of legacy data.
The principle of EIVs was independently invented by Heinz Ellenberg in Germany and L.G. Ramensky in Russia. Due to their high utility, to date, more than 30 EIV systems have been published in Europe, largely varying in indicators, definitions, scaling and plant nomenclature, thus impeding pan-European studies. To overcome these impediments, in early 2023, within a few days, two EIV systems were published for Europe: the Ellenberg-type indicator values by Tichý et al. (2023) and the Ecological Indicator Values for Europe (EIVE) 1.0 by Dengler et al. (2023). The new systems seem to match an urgent need, as both papers are within the top 1% most cited papers of the year 2023 according to the Scopus database.
With 14,835 valid taxa, EIVE is more comprehensive than Tichý et al. (2023) with 8,679 valid taxa, and it also has a larger spatial coverage (for a brief comparison of both systems, see https://vegsciblog.org/2023/01/21/eive-1-0/). Other than that, it was largely unknown which of the two systems performs better and how their performance relates to the performance of regional EIV systems. Only Dengler et al. (2023) contained correlations of species temperature indicators with GBIF-derived temperature niches, which indicated that EIVE performs slightly better than Tichý et al. (2023) and clearly better than most of the regional EIV systems.
While comparing different EIV systems became relevant only recently, the question of how to compute mean EIVs from the species’ EIVs was unresolved for ages. Both cover-weighted and unweighted means are widespread in the literature but without clear arguments, let alone empirical support for one of the solutions (see the review by Diekmann 2003). One could also think of an intermediate solution like square-root cover weighting. Recently, Hájek et al. (2020) proposed inverse niche-width weighting and found that, in certain scenarios, it outperforms other weighting approaches.
In this study, we used three regional datasets of vegetation plots combined with in-situ measured pH values and near-surface annual temperatures, respectively. We used the two European EIV systems (Dengler et al. 2023; Tichý et al. 2023) and the two regional EIV systems applicable for the Swiss Alps (Ellenberg et al. 1991; Landolt et al. 2010). We combined them with four different weighting approaches, namely unweighted (presence), square-root cover weighted, cover weighted and inverse niche-width weighted, the latter only being applicable to “EIVE” and “Landolt”. The performance of the different combinations was assessed via Pearson’s correlation coefficients (r) between mean EIV values and actual site conditions.
The first important observation was that – after taxonomic matching – only EIVE contained all valid taxa of the study, whether they were subspecies, species or aggregates, while the three other systems missed a significant number of valid taxa, either completely or by presenting them only at a higher or lower taxonomic level. In the latter cases, an approximative manual assignment would be, of course, possible, but it comes with additional work and arbitrariness. Moreover, while EIVE provides indicator values for all included taxa, just with different niche widths, the other three systems consider many taxa as indifferent and thus do not rate them. These aspects combined meant that dependent on the EIV system and the indicator, the three systems other than EIVE could not use between 12% and 40% of all occurring taxa for the calculation of mean indicator values.
When it comes to predicting site conditions, expectedly all four EIV systems can do that with only moderate differences in mean r values. However, when calculated with EIVE, the correlations were significantly better than when using “Tichý”. By contrast, “Ellenberg” and “Landolt” did not differ significantly from EIVE. Considering the weighting approach, no weighting performed significantly better than cover weighting, while square-root cover weighting was intermediate. In those two EIVE systems that provide niche-width information, no weighting and inverse niche-width weighting were equally good.
Our partly unexpected results might be explained by the “wisdom of the crowd” principle, according to which estimates averaged over several independent sources give better results than the assessment by a single good expert (Galton 1907; Surowiecki 2004). Accordingly, EIVE values based on 31 EIV systems should be better than Ellenberg-type indicator values, which are based on 12 EIV systems. Likewise, applying no cover-weighting means that effectively more taxa enter into the mean EIV value.
For the practice of vegetation ecologists in Europe, our study suggests that one should definitely not use full cover-weighting. EIVE or well-established regional EIV systems can be used, but the system of Tichý et al. (2023) is less advisable. Evidently, our study was based on three relatively small samples collected in the very centre of Europe and only for two indicators. It would be important to conduct similar “calibration” studies also in other parts of Europe (or across the entire continent) and for the other indicators. Finally, it is worth mentioning that currently the preparation of EIVE 1.5 is in the final stages, which will contain more than 20,000 valid taxa.
The idea for this paper originated from the initial work on a project conducted as part of the Swiss-Polish-Ukrainian Master Summer School “Biodiversity Monitoring” in Switzerland, during which students could learn about the vegetation ecology of alpine habitats, improve their understanding of statistical concepts, and admire the undeniable beauty of the Swiss Alps. During the 10 days spent in Preda, Switzerland, we sampled vegetation plots and analysed soil pH, which, together with data from previous conductances of the class, laid the foundation for this paper. Despite the relative lack of experience, working under proper supervision and applying newly acquired skills from the Summer School helped further develop this idea and turn it into a proper scientific article.
The statistical principle of the “wisdom of the crowd” suggests that a larger group of people can collectively make better decisions than a smaller one of a few experts. Involving as many researchers as possible in the scientific process, even inexperienced students or young researchers, can help to innovate and create new solutions.
Original study
Ostrowski G, Aicher S, Mankiewicz A, Chusova O, Dembicz I, Widmer S, Dengler J (2025) Mean ecological indicator values: use EIVE but no cover-weighting. Vegetation Classification and Survey 6: 57-67. https://doi.org/10.3897/VCS.134800
References:
- Dengler J, Jansen F, Chusova O, Hüllbusch E, Nobis MP, Van Meerbeek K, Axmanová I, Bruun HH, Chytrý M, … Gillet F (2023) Ecological Indicator Values for Europe (EIVE) 1.0. Vegetation Classification and Survey 4: 7–29. https://doi.org/10.3897/VCS.98324; see also https://vegsciblog.org/2023/01/21/eive-1-0/
- Diekmann M (2003) Species indicator values as an important tool in applied plant ecology – a review. Basic and Applied Ecology 4: 493–506.
- Ellenberg H, Weber HE, Düll R, Wirth V, Werner W, Paulißen D (1991) Zeigerwerte von Pflanzen in Mitteleuropa. Scripta Geobotanica 18: 1–248.
- Galton, F. (1907) Vox populi. Nature 75: 450–451.
- Hájek M, Dítě D, Horsáková V, Mikulášková E, Peterka T, Navrátilová J, Jiménez-Alfaro B, Tichý L, Horsák M (2020) Towards the pan-European bioindication system: Assessing and testing updated hydrological indicator values for vascular plants and bryophytes in mires. Ecological Indicators 116: 106527. https://doi.org/10.1016/j.ecolind.2020.106527
- Landolt E, Bäumler B, Erhardt A, Hegg O, Klötzli F, Lämmler W, Nobis M, Rudmann-Maurer K, Schweingruber FH, … Wohlgemuth T (2010) Flora indicativa – Ökologische Zeigerwerte und biologische Kennzeichen zur Flora der Schweiz und der Alpen. 2nd ed. Haupt, Bern, CH, 378 pp.
- Surowiecki, J. (2004) The wisdom of crowds. Doubleday, New York, US, 336 pp.
- Tichý L, Axmanová I, Dengler J, Guarino R, Jansen F, Midolo G, Nobis MP, Van Meerbeek K, Attorre F., … Chytrý M (2023) Ellenberg-type indicator values for European vascular plant species. Journal of Vegetation Science 34: e13168. https://doi.org/10.1111/jvs.13168