Ellenberg ’ s Indicator values for the Flora of Italy – first update : Pteridophyta , Gymnospermae and Monocotyledoneae

Guarino, R., Domina, G. & Pignatti, S.: Ellenberg’s Indicator values for the Flora of Italy – first update: Pteridophyta, Gymnospermae and Monocotyledoneae. — Fl. Medit. 22: 197-209. 2012. — ISSN: 1120-4052 printed, 2240-4538 online. Ellenberg’s indicator values are an useful tool to delineate the relationship between plants and environment, recognising to each species a functional role as biological indicator. In the frame of the second edition of the Pignatti’s “Flora d’Italia”, some new informative systems are under preparations, in order to support geobotanical/applied studies, including a complete and updated review of the Ellenberg’s indicator values for the whole bulk of species mentioned in the flora of Italy. This first contribution includes a list of 380 species of Pteridophyta, Gymnospermae and Monocotiledoneae that complete the first assignment of the Ellenberg’s indicator values to the flora of Italy, published in 2005. Besides, some methodological considerations on the attribution and the use of Ellenberg’s indicator values are reported.


Introduction
Ellenberg's indicator values (Eivs) have been proposed to estimate the influence of main environmental factors in determining flora and vegetation changes on a considered surface area (Ellenberg 1974(Ellenberg , 1996;;Ellenberg & al. 1992).Basing on a large observational and experimental evidence that ecological factors are the main determinant of structure and composition of plant communities, Ellenberg outlined the synecological preferences of all species belonging to the vascular flora of Central Europe, by means of numerical indices referring to 7 main environmental factors.These can be divided in two subgroups of three and four indices respectively: the first three indices refer to climatic variables: light conditions (L), temperatures (T), climatic continentality (C); the last four deal with edaphic conditions: moisture (U), reaction (R), nutrient availability (N), salinity (S).
All Eivs are arranged in ordinal scales, in which only a nominal correlation is given with the physical/chemical parameters measuring the environmental variables to which they are referred.In the original scales (Ellenberg & al. 1992), all indices were ranging between 1 and 9, with the only exceptions of U, ranging between 1 and 12 and S, ranging between 0 and 9.In more recent times, S-range have been limited to 0-3 (Ellenberg 1996;Pignatti & al. 2005).
From Central Europe, Eivs have been extended eastwards to Poland by Zarzycky (1984) and to Hungary by Borhidi (1995).This application was not so problematic, owing to the large number of species in common and to the comparable latitudinal range.In subsequent years, they were proposed for the flora of the south-Aegean Region (Böhling & al. 2002) and, for the first time, the opportunity to enlarge Ellenberg's scales in order to adapt them to warmer climatic conditions had to be discussed.Böhling & al. (2002) deemed not necessary to modify the Ellenberg's ordinal scales, considering that they are adimensional and, therefore, can be re-calibrated at regional scale in order to make them coherent with the ecological behaviour of local plant species.In this way, for instance, Quercus pubescens happens to have T = 6 for the Cretan flora and T = 8 for the German one.The reason is rather obvious: downy oak in Germany indicates relatively warm and dry environmental conditions, while in Crete it is linked to relatively mesic habitats.
If Ellenberg's scales are left unchanged when moving to different latitudes, ecological comparisons with different regional floras become senseless.This problem became very evident when was tried to extend the Eivs to the Italian territory.Even if the ecological conditions in northern Italy are comparable to those of southern Germany and a significant number of species are in common, this is not true for the southern and insular Italian territories.For this reason an enlargement of the scales L and T to 12 values, in order to respect the statistic homogeneity with the original Ellenberg's assumption was proposed (Pignatti & al. 2005).
The Eivs for the Italian flora published by Pignatti & al. (2005) were based on the species list of the first edition of the renowned "Flora d'Italia" (Pignatti 1982), whose second edition is expected to be published (...hopefully!) by the end of the next year.Aim of this paper is to provide a first integration of the Eivs for the Italian Flora and, at the same time, to offer some methodological considerations on the attribution and the use of Ellenberg's indicator values.
The difference from the species list of the first edition consists of 1518 species, of which 380 are pteridophytes, gymnosperms and monocots.These groups are the subject of our first contribution, where species are ordered according to APG (2006) and to Christenhustz & al. (2011aChristenhustz & al. ( , 2011b) ) for pteridophytes and gymnosperms.
In order to assign the new Eivs, all the criteria and recommendations adopted by Ellenberg & al. (1992) have been followed, even if the criterion "per analogy" was leading our choices in most of the cases: given the growing stands and sociological behaviour of species a, the attribution of its Eivs followed in most of the cases the Eivs already assigned to the most frequent and abundant species in its growing stands or, in case of ambiguous magnitudes and correlations, the Eivs of species a was arbitrarily assigned (basing on our experience) within the confidence interval delimited by the Eivs already assigned to the most frequent and abundant species in its growing stands.

Results
Synecological requirements of a species may change along its distribution range, especially when moving to different latitudes.In order to extend the Eivs to the Italian flora, a data-base was implemented by Pignatti and his collaborators since more than twenty years.In such data-base, all the species of the first edition of "Flora d'Italia" were reported, as well as the phytosociological frame and results of ecological and ecophysiological measurements as far as available for each species.So far, that data-base was incremented with 380 newly assigned Eivs, which are reported in Annex 1.A second contribution, which will include the dicots still missing is currently under preparation.
On the website of the Herbarium Mediterraneum Panormitanum where Flora Mediterranea is published (http://www.herbmedit.org), the complete list of Eivs for the flora if Italy will be soon made available in the form of a Microsoft Excel ® spreadsheet, as Electronic supplementary file in order to be easily imported in data-bases or readily used for data processing.This is an optimal condition for open-ended works, like floras and any kind of related data-base are per definition.The digital list of the Eivs will be periodically updated and adjusted, following the indications of all users, in order to progressively improve their accuracy.As already noted by Pignatti & al. (2005) even the current list of Eivs has to be considered a first approximation, which may require a long period of adjustments and revisions.Ellenberg himself, when proposing the first edition of his indicator values, stated that it was a work in progress (Ellenberg 1974).The same was repeated in the following editions, defined as "further approximations".
While assigning the new Eivs, we reviewed, as well, the distribution frequency of all values in the list and we noticed that the value L = 10 was critically underrepresented in the list of Pignatti & al. (2005).So, we reconsidered the pristine criterion of attributing L = 11 to most of the Mediterranean species and assigned the value L = 10 to many synanthropic species and to many small sized annuals commonly found in Mediterranean garigues and perennial dry grasslands.This is just one example of how Eivs can change and may further change in the future, following new evidences.
Other examples of Eivs that have been modified from the edition of 2005 are the following: -Arundo plinii: due to the re-evaluation of Arundo collina, the former species should be limited to the loamy and clayish riverbeds of north-eastern Italy: Marche, Emilia-Romagna and Friuli-Venezia Giulia.For this reason, its Eivs have been modified as follows: C = 6, U = 6, R = 5, N = 5 -Iris pseudopumila: R = 6, instead of R = 4 -Limodorum abortivum: L = 4, instead of L = X

Discussion
An important feature of Ellenberg's indices is that they do not express the physiological optimum of a given species, but to its synecological optimum, that is the optimal ecological requirements of a given species when interacting with other species.These interaction can vary from region to region and even from time to time.An approach to assess the significance of observed variations (in space and time) of Eivs has been proposed by Pignatti & al. (2001a).
One might ask what is the meaning of Eivs in the era of environmental monitoring sensors and weatherproof digital loggers, which can be used for direct measurements of a large number of variables at any given time-interval.As already stated by Ellenberg himself, the indicator values do not replace measurements.Instead, they should be seen as practical tools to draft quick estimates for eco-logical interpretations of plant-species lists.On the other hand, they may help to design the distribution of a network of data loggers to corroborate and integrate with experimental evidence the indicator values.A bit like the interpolations of climate parameters: the higher is the number of meteorological stations recording real data to set and train models, the better is the accuracy and performance of predictive systems.
Some authors (Durwen 1982;Böcker & al. 1983;Kowarik & Seidling 1989;Möller 1992), even if sometimes recognising a practical utility in calculating average Eivs of a species list, state that this is not mathematically correct, because Eivs are ordinal scales, without dimensional correlation with chemical/physical parameters.It is well demonstrated (see, for instance, Pignatti & al. 2001a) that when the number samples (i.e.specie lists) is high enough, Eivs do fit the normal distribution, whose essential parameters are average and variance.So, in addition to non-parametric statistics, all kinds of statistical tests based on average and variance can be used.If only a small number of samples is available, a parametric statistic approach can be essayed anyway, because in the case of Ellenberg's scales it is always possible to transform data in their respective ranks, to better approximate the normal distribution and to create the conditions to use a parametric approach.
Moreover, several contributors demonstrated a fairly good correlation between Eivs and dimensional measurements of environmental variables (Degorski 1982;Seidling & Rohner 1993;van der Maarel 1993;Möller 1997;Petersen 2000;Schaffers & Sýkora 2000;Onori 2002).Thus, to perform parametric statistic tests and to use average indicator values referred to floras and plant communities should be considered always admissible, as already accepted by many authors.This greatly enlarges the utility of such indices.van der Maarel, E. 1993 Annex 1. Eivs assigned to the considered species.The value "X" stands for "undetermined" i.e. species with a very wide ecological amplitude for the considered variable.The value "0" stands for "unknown" (i.e.data deficient), with the exception of S, where "0" means "no salinity".Additional informations on the meaning of Eivs can be found in Pignatti & al. (2005).