Bias in Ellenberg indicator values — problems with detection of the effect of vegetation type

A recent analysis published in this journal found different relationships between mean Ellenberg indicator values and environmental measurements in different vegetation types. The cause was stated as bias in mean Ellenberg values between relevés which in turn suggested to reflect a bias in individual Ellenberg values. We discuss two phenomena that could explain these results without the need to invoke bias in either individual or mean Ellenberg values. Firstly, slopes of linear regression lines underestimate true relationships when analyses involve explanatory variables measured with error. Secondly, syntaxon‐specific distributions of Ellenberg values follow from the floristic definition of phytosociological units. Mean Ellenberg values per relevé therefore carry the stamp of their associated syntaxon even though associated abiotic conditions may vary between relevés. This will lead to variation in slopes and intercepts between vegetation types not because of bias in individual Ellenberg values but because of prescribed bias in the distribution of Ellenberg values between syntaxa. The residual variation in calibrations carried out across vegetation types is undoubtedly reduced by introducing vegetation type as a factor. However users should note that this is unlikely to reflect bias in individual Ellenberg values but is more likely to reflect error in environmental measurements as well the constraint imposed by phytosociological classification.     

Do we really need phytosociological classes to calibrate Ellenberg indicator values?

Abstract. Wamelink et al. (2002) calibrated Ellenberg indicator values for acidity and water availability against measured soil pH and measured mean spring groundwater level (MSL), respectively. Linear regression between indicator value and measured value of all the observations gave a poor fit. Regression lines per phytosociological vegetation class, on the other hand, generally described the observations well. In this article we demonstrate that this result is, at least partly, an artefact. First, because the data utilized are likely to contain systematic errors, and second, because a wrong regression model was applied. A sigmoid function for the relation between the indicator value for water availability and MSL gives a far better fit than a linear function does.‘Vegetation class’ is not an obvious choice as an extra explanatory variable for the regression, as it is only a convenient label for vegetation and should not be used as if it were a real independent environmental variable. In general, indicator values of plant species should be calibrated against environmental variables with great care. This implies that researchers should have knowledge about the ecological demands plants make on their environment, as well as about the spatial and temporal variability of this environment.     

Vegetation of beech forests in the Rychlebské Mountains,Czech Republic,re-inspected after 60 years with assessment of environmental changes

From 1941–;1944 nearly 30 phytosociological relevés were completed by F. K. Hartmann in the Rychlebské Mountains, a typical mountainous area in northeastern Czech Republic. Of the original plots still covered with adult grown beech (Fagus sylvatica) forest, 22 were resampled in 1998 and 1999. In order to describe the recent vegetation variability of the sites 57 relevés were recorded. Changes in vegetation were estimated using relative changes in species density and ordinations (PCA, RDA). Environmental changes were assessed using Ellenberg indicator values when no direct measurements were available. A decline in species diversity has been documented, particularly, many species occurring frequently in deciduous forests with nutrient and moisture well-supplied soils around neutral have decreased. In contrast, several light-demanding, acid- and soil desiccation-tolerant species have increased. Natural succession, quantified as forest age, contributed slightly to these changes. In Ellenberg indicator values, a decline in F (soil moisture), R (soil calcium) and N (ecosystem productivity), and an increase in L (understorey light) were shown. This is interpreted as the influence of modified forestry management and of airborne pollutants. Intensified logging caused the canopy to open and soil conditions to worsen. The latter is most likely also due to acid leaching of soil cations (Ca, K, Na). This caused a decline in soil productivity, thus the effect of nitrification could not be detected. The original relevés may have differed in size influencing the results. This revised version was published online in August 2006 with corrections to the Cover Date.     

Bayesian classification of vegetation types with Gaussian mixture density fitting to indicator values

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 reléves 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.     

Ecological preferences of alien plant species in North-Eastern Germany

The large, comprehensive vegetation database of Mecklenburg-Vorpommern/NE Germany with 51,328 relevés allowed us to study an entire regional flora of 133 non-native plants (NNP, immigration after 1492 AD) with regard to their preferences to all kinds of habitats and along different ecological gradients. For each relevé, we computed average Ellenberg indicator values (EIV) for temperature, light, moisture, reaction, nutrients and salt as well as plant strategy type weights. We partitioned the dataset into relevés with and without occurrences of NNP and compared them with respect to the relative frequencies of EIVs and strategy type weights. We identified deviations from random differences by testing against permuted indicator values. To account for bias in EIV between community types, NNP preferences were differentiated for 34 phytosociological classes. We tested significance of preferences for the group of NNP as a whole, as well as for single NNP species within the entire dataset, as well as differentiated by phytosociological classes and formations. NNP as a group prefer communities with high EIVs for temperature and nutrients and low EIVs for moisture. They avoid communities with low EIV for reaction and high EIV for salt. NNP prefer communities with high proportions of ruderal and low proportion of stress strategists. The differentiation by phytosociological classes reinforces the general trends for temperature, nutrients, moisture, R and S strategy types. Nevertheless, preferences of single species reveal that NNP are not a congruent group but show individualistic ecological preferences.     

Trends in species diversity and composition of urban vegetation over three decades

Question: What was the change in diversity of urban synantropic vegetation in a medium‐sized Central European city during the period of increasing urbanization (1960s‐1990s)? Location: The city of Plzeň, an industrial centre of the western part of the Czech Republic. Methods: Sampling of various types of synanthropic vegetation, conducted in the 1960s, was repeated by using the same methods in the 1990s. This yielded 959 relevés, of which 623 were made in the 1960s and 336 in the 1990s. The relevés were assigned to the following phytosociological classes: Chenopodietea, Artemisietea vulgaris, Galio‐Urticetea, Agropyretea repentis and Plantaginetea majoris. Total number of vascular plant species, evenness index J, number of alien species (classified into archaeophytes and neophytes), and mean Ellenberg indicator values for light, temperature, continentality, moisture, soil reaction, and nutrients were obtained for each relevé. Results: From 1960s to 1990s, there was a significant decrease of species richness and diversity in synanthropic vegetation. The proportion of archaeophytes decreased in most vegetation types, indicating the contribution of this group of species, often confined to specific rural‐like habitats, to the observed impoverishment of ruderal vegetation. The proportion of neophytes did not change between the two periods. Comparison between 1960s and 1990s indicated a decrease in light, temperature, moisture, soil reaction and nutrient indicator values in some vegetation types. In both periods, Artemisieta, Galio‐Urticetea and Chenopodietea formed a distinct group harbouring more species than Agropyretea and Plantaginetea. Neophytes, i.e. recently introduced species, were most represented in the early successional annual vegetation of Chenopodietea, rather than in perennial vegetation of the other classes. Conclusions: Synanthropic vegetation of Plzeň exhibited a general trend of decrease in species diversity.     

Vegetational changes in a wet meadow complex,South Bohemia,Czech Republic

Directional changes of vegetation in a wet meadow complex are described, comparing phytosociological relevés taken in 1956, 1963, 1984 and 1989, and vegetation maps made in 1956 and 1984. The relevés were located at various distances from the pond shore predominantly reflecting differences in water regime and intensity of farming. The data were elaborated using direct and indirect gradient analyses. Alterations of environmental factors (moisture, nitrogen) were expressed using Ellenberg's indicator values. Over most of the area, the former distinct vegetation mosaic has been replaced by uniform meadows due to intensive farming. The area was eutrophicated and partly drained, clearly shown in Ellenberg's indicator values, in species composition and vegetation pattern. Rapid processes of ruderalization took place. A scheme of successional pathways which finally emerged can be applied to analogous wet meadows in comparable geographical regions and used as a tool for prognoses of successional processes in meadows under human impact. The paper demonstrates ways to exploit earlier phytosociological records.     

On the statistical analysis of vegetation change: a wetland affected by water extraction and soil acidification

Abstract. A case study is presented on the statistical analysis and interpretation of vegetation change in a wetland subjected to water extraction and acidification, without precise information on the environmental changes. The vegetation is a Junco-Molinion grassland and the changes in vegetation are evaluated on the basis of relevés in 1977 and 1988 of 20 plots in a small nature reserve on moist oligotrophic, Pleistocene sands in the Netherlands. The changes are attributed to water extraction (since 1972) and soil acidification and the effect of the environmental changes on the vegetation is inferred from data on water depth and acidity collected in 1988. Many species typical of wetlands decreased in abundance, including rare species such as Parnassia palustris, Selinum carvifolia and Ophioglossum vulgatum. Some species increased, notably Anthoxanthum odoratum, Holcus lanatus and Plantago lanceolata. A significant decrease was found in the mean Ellenberg indicator values for moisture and acidity. The mean indicator value for nutrients did not change significantly. Multivariate analysis of the species data by Redundancy Analysis demonstrated the overall significance of the change in species composition between 1977 and 1988 (P < 0.01, Monte Carlo permutation). The spatial and temporal variation in the species data was displayed in ordination diagrams and interpreted in terms of water depth and pH. A simple model is developed to infer the change in water depth and pH from the relevé data and recent data on water depth and pH. Because the correlation between water depth and pH made a joint estimation of the changes useless, the change in pH was estimated for a series of likely changes in water depth. For the most likely change in water depth, significant acidification was inferred from the change in vegetation. The model is more generally applicable as a constrained calibration method.     

An application case of ecological indicator values (Zeigerwerte) calculated with a simple algorithmic approach

Abstract

The vegetation of the study site near Rome (Castelporziano Estate), where different woodland types occur, was analysed on the basis of ecological indicator values (Zeigerwerte) for light, temperature, continentality of climate, soil moisture, soil pH and nitrogen. Indicator values were estimated with Hill's reprediction algorithm for the flora of Central-Southern Italy relying on a database of 4,207 original relevés representing a balanced survey of the vegetation of this and surrounding areas. It was possible to obtain indicator values for an important fraction of the Italian Mediterranean flora. Results are ecologically reasonable, and it was possible to find strong correlation between the recalculated values and a few environmental variables. These correlations were not significant in an analogous test with subjectively derived scores of Ellenberg indicator values.     

Cover‐weighted averaging of indicator values in vegetation analyses

Question: How should species cover be weighted when calculating average indicator values of vegetation relevés? Location: The Netherlands. Method: Various weighting methods were statistically investigated with 188 relevés from The Netherlands for which accurate groundwater levels were available. For each method the correlation between average Ellenberg indicator value for moisture and mean spring groundwater level was calculated. A permutation test on correlation coefficients revealed whether differences between methods were significant or not. Results: Optimization of a general weighting function did not produce a significantly higher correlation than disregarding cover and calculating the average as the arithmetical mean of indicator values. Giving a higher weight to species at both ends of the indicator scale and using indifferent species as indicators of mediocre conditions did improve the correlation significantly. Weighting species proportionate to their cover yielded a significantly lower correlation than the correlation obtained with the method that disregards cover. A significantly lower correlation was also established when taking into account the fact that cover is related to the growth strategy of species.     

Calibrating Ellenberg indicator values for moisture,acidity, nutrient availability and salinity in the Netherlands

A general calibration of Ellenberg indicator values for moisture, acidity, nutrient availability and salinity was carried out on a large database of relevées and environmental variables from a variety of ecosystems in the Netherlands.Satisfying relationships with Ellenberg indicator values for moisture, acidity and salinity were found for mean groundwater level in spring time, soil pH and chloride concentration in groundwater. For mean groundwater level in spring and chloride concentration in groundwater subdivision of the database led to clearer relationships with indicator values. For the Ellenberg indicator value for nutrient availability satisfying calibration results were only achieved with data on standing crops and N stock in standing crop. The relationship with soil chemical variables was less clear.Although the correlation between indicator and measured values is obvious, the variation around the regression lines is considerable. However, because of the size and composition of the database, it is unlikely that our calibration results can be much improved by adding more (Dutch) data.The calibration results will be applied in the multi-stress model SMART-MOVE, developed to predict changes in species composition due to acidification, eutrophication and the effects of lowering groundwater.     

Long‐term impacts of nitrogen and sulphur deposition on forest floor vegetation in the Northern limestone Alps,Austria

Question: Are there effects of long‐term deposition of airborne nitrogen and sulphur on the forest floor vegetation from permanent plots collected in 1993 compared to 2005. Location: Northern limestone Alps in Austria. Methods: Single species responses were analysed by correlating trends in cover‐abundance values, as derived from marginal models, with Ellenberg indicator values. Changes in the species composition of plots were analysed by correlating changes in mean Ellenberg indicator values with the displacement of plots within a multidimensional scaling ordination. Results: Trends in single species abundance were positively correlated with indicator values of soil pH but were independent of nutrient availability. A general trend towards the homogenisation of vegetation, due to convergent time vectors of the relevés, became obvious. Oligotrophic sites previously situated at the distal ends of ordination axes shifted towards the centre since they were enriched by species preferring mesotrophic conditions. The bulk of plots with intermediate site conditions hardly showed any trends. A concomitant analysis demonstrated that temporal changes in species composition exceed the variation in cover abundance estimates among different field botanists. Conclusions: N deposition can lead to a homogenisation of forest floor vegetation. Larger limestone areas with diverse soil conditions, such as the Northern limestone Alps in Austria, as a whole are thus negatively affected by airborne N deposition. Nevertheless, the vegetation was at least as strongly affected by an increase of basiphilous species as a result of decreasing S deposition.     

Phytosociological data give biased estimates of species richness

Abstract. Large phytosociological data sets of three types of grassland and three types of forest vegetation from the Czech Republic were analysed with a focus on plot size used in phytosociological sampling and on the species‐area relationship. The data sets included 12975 relevés, sampled by different authors in different parts of the country between 1922 and 1999. It was shown that in the grassland data sets, the relevés sampled before the 1960s tended to have a larger plot size than the relevés made later on. No temporal variation in plot sizes used was detected in forest relevés. Species‐area curves fitted to the data showed unnatural shapes, with levelling‐off or even decrease in plot sizes higher than average. This distortion is explained by the subjective, preferential method of field sampling used in phytosociology. When making relevés in species‐poor vegetation, researchers probably tend to use larger plots in order to include more species. The reason for this may be that a higher number of species gives a higher probability of including presumed diagnostic species, so that the relevé can be more easily classified in the Braun‐Blanquet classification system. This attitude of phytosociologists has at least two consequences: (1) in phytosociological data bases species‐poor vegetation types are underrepresented or relevés are artificially biased towards higher species richness; (2) the suitability of phytosociological data for species richness estimation is severely limited.     

Using species‐environmental amplitudes to predict pH values from vegetation

Question: Species optima or indicator values are frequently used to predict environmental variables from species composition. The present study focuses on the question whether predictions can be improved by using species environmental amplitudes instead of single values representing species optima. Location: Semi‐natural, deciduous hardwood forests of northwestern Germany. Methods: Based on a data set of 558 relevés, species responses (presence/absence) to pH were modelled with Huisman‐Olff‐Fresco (HOF) regression models. Species amplitudes were derived from response curves using three different methods. To predict the pH from vegetation, a maximum amplitude overlap method was applied. For comparison, predictions resulting from several established methods, i. e. maximum likelihood/present and absent species, maximum likelihood/present species only, mean weighted averages and mean Ellenberg indicator values were calculated. The predictive success (squared Pearson's r and root mean square error of prediction) was evaluated using an independent data set of 151 relevés. Results: Predictions based upon amplitudes defined by maximum Cohen's x probability threshold yield the best results of all amplitude definitions (R²= 0.75, RMSEP = 0.52). Provided there is an even distribution of the environmental variable, amplitudes defined by predicted probability exceeding prevalence are also suitable (R²= 0.76, RMSEP = 0.55). The prediction success is comparable to maximum likelihood (present species only) and – after rescaling – to mean weighted averages. Predicted values show a good linearity to observed pH values as opposed to a curvilinear relationship of mean Ellenberg indicator values. Transformation or rescaling of the predicted values is not required. Conclusions: Species amplitudes given by a minimum and maximum boundary for each species can be used to efficiently predict environmental variables from species composition. The predictive success is superior to mean Ellenberg indicator values and comparable to mean indicator values based on species weighted averages.     

JUICE,software for vegetation classification

Abstract. The program JUICE was designed as a Microsoft® WINDOWS® application for editing, classification and analysis of large phytosociological tables and databases. This software, with a current maximum capacity of 30 000 relevés in one table, includes many functions for easy manipulation of table and header data. Various options include classification using COCKTAIL and TWINSPAN methods, calculation of interspecific associations, fidelity measures, average Ellenberg indicator values, preparation of synoptic tables, automatic sorting of relevé tables, and export of table data into other applications (word processors, spreadsheet programs or mapping packages). JUICE is optimized for use in association with TURBOVEG which is the most widespread database program for storing phytosociological data in Europe.     

Empirical realised niche models for British higher and lower plants – development and preliminary testing

Question: Can useful realised niche models be constructed for British plant species using climate, canopy height and mean Ellenberg indices as explanatory variables? Location: Great Britain. Methods: Generalised linear models were constructed using occurrence data covering all major natural and semi‐natural vegetation types (n=40 683 quadrat samples). Paired species and soil records were only available for 4% of the training data (n=1033) so modelling was carried out in two stages. First, multiple regression was used to express mean Ellenberg values for moisture, pH and fertility, in terms of direct soil measurements. Next, species presence/absence was modelled using mean indicator scores, cover‐weighted canopy height, three climate variables and interactions between these factors, but correcting for the presence of each target species in training plots to avoid circularity. Results: Eight hundred and three higher plants and 327 bryophytes were modelled. Thirteen per cent of the niche models for higher plants were tested against an independent survey dataset not used to build the models. Models performed better when predictions were based only on indices derived from the species composition of each plot rather than measured soil variables. This reflects the high variation in vegetation indices that was not explained by the measured soil variables. Conclusions: The models should be used to estimate expected habitat suitability rather than to predict species presence. Least uncertainty also attaches to their use as risk assessment and monitoring tools on nature reserves because they can be solved using mean environmental indicators calculated from the existing species composition, with or without climate data.     

Reliability of Ellenberg indicator values for moisture,nitrogen and soil reaction: a comparison with field measurements

Abstract. Ellenberg indicator values for moisture, nitrogen and soil reaction were correlated with measured soil and vegetation parameters. Relationships were studied through between‐species and between‐site comparisons, using data from 74 roadside plots in 14 different plant communities in The Netherlands forming a wide range. Ellenberg moisture values correlated best with the average lowest moisture contents in summer. Correlations with the annual average groundwater level and the average spring level were also good. Ellenberg N‐values appeared to be only weakly correlated with soil parameters, including N‐mineralization and available mineral N. Instead, there was a strong relation with biomass production. We therefore endorse Hill & Carey's (1997) suggestion that the term N‐values be replaced by ‘productivity values'. For soil reaction, many species values appeared to need regional adjustment. The relationship with soil pH was unsatisfactory; mean indicator values were similar for all sites at pH > 4.75 because of wide species tolerances for intermediate pH levels. Site mean reaction values correlated best (r up to 0.92) with the total amount of calcium (exchangeable Ca²⁺ plus Ca from carbonates). It is therefore suggested that reaction values are better referred to as ‘calcium values'. Using abundance values as weights when calculating mean indicator values generally improved the results, but, over the wide range of conditions studied, differences were small. Indicator values for bryophytes appeared well in line with those for vascular plants. It was noted that the frequency distributions of indicator values are quite uneven. This creates a tendency for site mean values to converge to the value most common in the regional species pool. Although the effect on overall correlations is small, relationships tended to be less linear. Uneven distributions also cause the site mean indicator values at which species have their optimum to deviate from the actual Ellenberg values of these species. Suggestions for improvements are made. It is concluded that the Ellenberg indicator system provides a very valuable tool for habitat calibration, provided the appropriate parameters are considered.     

Measurement errors and regression to the mean cannot explain bias in average Ellenberg indicator values

Smart & Scott (2004, this is sue) criticized our paper (Wamelink et al. 2002) about the bias in average Ellenberg indicator values. Their main criticism concerns the method we used, regression analysis. They state the bias can be mimicked by the construction of an artificial data set and that regression analysis is not a suited tool to investigate underlying phenomena. Moreover they claim that the present bias is caused by the distribution of Ellenberg indicator values between syntaxa, instead of a bias in average Ellenberg indicator values per species. We show that their criticism of the use of regression analysis does not hold. We selected average Ellenberg values per vegetation group for several pH classes and applied an F‐test to determine whether or not the vegetation groups within each pH class differed significantly from each other. This was the case for all tested classes (P < 0.001). Moreover we simulated an artificial data set, of which the F‐test for varying measurement error could not explain the magnitude of the F‐value we found earlier. This indicates that the bias we found in average Ellenberg indicator values cannot be explained by measurement errors or by regression to the mean. In the end, Smart & Scott, as we did, come to the conclusion that there is a bias present and that separate regression lines per vegetation type are necessary, but the debate remains open on whether or not this is caused by the bias in Ellenberg indicator values per species.     

Indicating Soil Acidity Using Vegetation Relevés in Spatially Limited Areas – Case Study from the Považský Inovec, Slovakia

A fine-scaled approach for predicting soil acidity using plant species in a spatially limited area (?epú?ky Nature Reserve, Slovakia) is presented here. This approach copes with some specific limitations: i) a limited pool of vegetation data may make the predictions too sensitive to the lack of species information, and ii) the predictions may be sensitive to the narrow pH gradient. Vegetation relevés and soil reaction (pH-H₂O and pH-CaCl₂) were systematically recorded. A set of species indicator values and amplitudes was calibrated with physical pH data using the Weighted Averaging (WA), HOF modelling and Non-Metric Multidimensional Scaling (NMDS) methods, along with Ellenberg indicator values. Two prediction methods were tested: i) WA and ii) Amplitude Overlap (AO). WA prediction with Ellenberg’s and WA-calibrated species indicator values were the most powerful technique (R ²?=?68.4–68.7% and 53.4–59.1% for pH-CaCl₂ and pH-H₂O, respectively). WA-prediction with HOF-based indicator values was less effective (R ²?=?61.7% and 50.7%) due to the decrease in species’ information because with HOF modelling many species are assumed indifferent or too rare. The NMDS method does not bring any significant gain to the calibration, though it avoids the lack of species information. The AO method was proven to be less powerful under studied circumstances, because it is sensitive both to the lack of species’ information and to the truncation of species responses. The results prove that a spatially explicit approach can provide significant indices to estimate changes in soil acidity – pH-CaCl₂ better than pH-H₂O.