Folia Geobotanica 33: 113-131, 1998 VEGETATION OF TUSCAN ULTRA MAFIC SOILS IN RELATION TO EDAPHIC AND PHYSICAL FACTORS Alessandro Chiarucci1^, Brett H. Robinson2'3\ Ilaria Bonini1^, Daniel Petit3 Robert R. Brooks ' & Vincenzo De Dominicis 1) Dipartimento di Biologia Ambientale, Universita degli Studi di Siena, Via PA. Mattioli 4, 1-53100 Siena, Italy; fax +39 577 298860, E-mail chiarucci@unisi.it 2) Department of Soil Science, Massey University, Private bag 11222, Palmerston North, New Zealand; fax +64 6 350 5632, E-mail r.brooks@massey.ac.nz 3) Laboratoire de Génétique et Evolution, CNRS, Universitě de Lille, F-59655 Vdleneuve d'Ascq, France Keywords: Canonical correspondence analysis, Drought, Serpentine, Soil features, Species diversity, Vegetation ecology Abstract: Vegetation and soil sampling were carried out in 80 plots located in five different ultramafic (serpentine) sites of Tuscany, central Italy. The physical and chemical features of each plot were determined and the species composition and cover recorded. The exchangeable fraction of soil metals was analysed because it gives a measure of their concentrations available to plants. The plots were classified by cluster analysis and ANOVA was used to compare the environmental variables of the groups of plots. Canonical correspondence analysis was used to detect the principal factors for gradients of species composition within the plant communities. A higher content of exchangeable metals was found under the more evolved and structured plant communities, suggesting that serpentine vegetation of Tuscany is not strongly limited by soil metals, such as chromium, cobalt, nickel and magnesium, typically associated with ultramafic soils. The low nutrient content of the soils and drought stress mainly due to topographical features, appear to have a more significant role in determining the typical scattered vegetation of the Tuscan ultramafics. INTRODUCTION Ultramafic (serpentine) soils host a distinctive flora and vegetation. The stunted plants and patches of bare soil have been attributed to the chemical and physical features of ultramafic soils. These typically contain high concentrations of heavy metals such as nickel, chromium, cobalt, and manganese, low concentrations of nutrients and a large excess of magnesium over calcium; moreover these soils are shallow, well drained and prone to erosion (Brooks 1987, Baker et al. 1992, Roberts & Proctor 1992). However, large variations exist in the soil features of the different sites. Proctor & Nagy (1992) suggested that the factors controlling ultramafic flora might differ from site to site. There are several ultramafic outcrops in Tuscany, Italy, and their flora and vegetation have been widely studied by botanists and ecologists (see Vergnano Gambi 1992). The most typical plant community of Tuscan ultramafic soils is a garigue, described as the Armerio-Alyssetum bertolonii association, which is characterised by low ground cover and several endemic species (Arrigoni et al. 1983, Chiarucci et al. 1995). This vegetation type occurs, differentiated into two sub-associations, over all ultramafic outcrops of Tuscany (Chiarucci et al. 1995). Tall scrub or woodland plant communities have also been reported 114 A. Chiarucci et al. Fig. 1. Distribution of the ultramafic outcrops (black) of Corsica and central-northern Italy with indications of study areas. A - Livorno hills; B - Cecina Valley and surrounding areas; C - Murlo area; D - Mt. Ferrato; e - Upper Tiber Valley. and are widespread in some sites (Messeri 1936, Pichi-Sermolu 1948, Marchiori & Tornadore Marchiori 1977, Chiarucci 1994). Some authors (e.g. Sasse 1979a, b, Arrigoni et al. 1983, Vergnano Gambi 1992) have claimed that soil metal content negatively affects plant communities. Others have hypothesised that soil metal content does not affect vegetation when nutrients are available (Chiarucci & De Dominicis 1995, Chiarucci 1996). In spite of the several papers published on the flora, vegetation and metal uptake by plants, only a few studies have thoroughly investigated the relations between environmental features and vegetation composition and structure (Chiarucci et al. 1998). Vegetation of Tuscan ultramafic soils 115 The aim of the present paper was to describe the species diversity and composition of the plant communities growing over the most important ultramafic outcrops of Tuscany in relation to the main environmental features, with particular attention to the exchangeable soil metal content. MATERIALS AND METHODS Study areas and vegetation sampling Five of the largest ultramafic areas in Tuscany were selected as study sites (Fig. 1): the hills near Livorno, the Cecina Valley and surrounding areas, the Murlo area, Mt. Ferrato and the upper Tiber Valley. Some 15-20 100 m2 plots, representing all the main vegetation types, were sampled in each area in June 1996. The cover of all vascular species was estimated in each vegetation layer. The following features were recorded: altitude (m), aspect, slope (°), cover (%) and height (m) of the tree layer, cover (%) and height (m) of the shrub layer, field layer cover (%), rockiness (%) and stoniness (%). The index of normalised insolation (Bartorelli 1967) was used as a heat index, summarising the interactive effects of latitude, slope and aspect on determining the amount of solar radiation that each plot receives during the year. This index based on the hours per year of overhead sunlight, is a measure of the radiation the site actually receives in one year, given its aspect, slope, and latitude. Plant nomenclature follows Pignatti (1982) for most species, Chiarucci et al. (1995) for serpentinophytes and Moraldo (1986) for the genus Stipa. Soil sampling and analysis A composite soil sample was taken in each plot at about 10 cm of depth. Soil samples were air dried and passed through a 2 mm sieve. A 1 : 2.5 soil-water suspension was used to measure pH; organic matter was estimated by measuring loss on ignition (LOI) after heating soil subsamples to 450 °C overnight. Exchangeable cations were determined by adding 30 ml of 1M CH3COONH4 to 3 g of soil in a polythene container and shaking the containers overnight. The solutions were filtered and analysed by flame atomic absorption spectroscopy. The exchangeable fraction, as opposed to the total metal concentration, was determined because it gives a measure of the concentration available to plants. Data analysis Because of the climatic and phytogeographic differences between the Mediterranean (the first three sites) and the inland (Mt. Ferrato and Upper Tiber Valley) sites (Arrigoni 1980, Chiarucci et al. 1995), the data-set was divided in two parts. The total cover of each species (not subdivided into three layers) was used for analysis. The plots were classified by a homogeneity-optimising procedure. Cluster analysis, using MNSSQ as agglomeration criterion, was applied to a distance matrix constructed on the basis of the similarity ratio (Podani 1994). Data processing was performed with the SYN-TAX 5.0 program package (Podani 1993). For the groups of plots identified by cluster analysis, soil and environmental data were compared by one-way ANOVA; statistical differences between groups were tested by the Duncan multiple-range test at P < 0.05. To satisfy the normality assumptions of statistical analysis, the variables LOI, rockiness, stoniness, and exchangeable cations were logarithmically transformed. 116 A. Chiarucci et al. Table 1. Structural features and species diversity data (mean ± s.e.) of plots belonging to the six groups identified by cluster analysis in the Mediterranean data-set. Cluster/Group n a/1 9 a/2 8 b/3 7 b/4 7 C/5 7 C/6 12 Tree layer cover (%) _ 11.4+ 11.4 55.0 ± 12.0 Shrub layer cover (%) 3.8 + 1.5 12.8 + 2.2 70.7 ± 5.2 83.6 ±2.6 73.4 ± 12.2 51.6 ±9.8 Field layer cover (%) 13.8 ± 2.1 14.0 + 2.4 25.0 ±4.4 17.7 + 6.3 15.7 ±9.6 2.1 ±0.5 Total species richhess 32.5 ± 1.2 29.6 ± 2.2 26.4 ± 2.3 26.4+1.7 26.6 ±2.0 15.9 + 1.1 Tree layer richness 0.0 + 0.0 0.0 + 0.0 0.0 ±0.0 0.0 ±0.0 0.9 + 0.9 3.3 ±0.8 Shrub layer richness 3.2 ±0.6 4.7 ±0.6 9.6 ± 1.0 13.0 ±0.8 11.1+0.9 8.8 ±0.8 Field layer richness 29.7 ± 1.5 25.0 ±2.1 16.9 ±2.3 13.6 ± 1.4 15.0 ±2.6 ' 5.6 ±0.7 To detect gradients in species composition and in species-environment relations, canonical correspondence analysis (CCA) (ter Braak 1986) was performed by the program package CANOCO 3.11 (ter Braak 1988). Cover data were log-transformed and rare species were downweighted. A Monte Carlo permutation test was applied to test the significance of the eigenvalue corresponding to the first CCA canonical axis (ter Braak 1988). RESULTS 191 vascular plant species were found in the eighty 100 m2 plots, of which 142 species were in the Mediterranean data set and 127 in the inland data set. The average species richness per plot was 25.1. The results are reported separately for the two data sets. Mediterranean data set Species composition and richness The main cluster divisions (not shown) were linked to floristic characters reflecting vegetation structure and diversity (Tab. 1). All groups contained plots from the three sites. Species frequency and cover in the six groups are reported in Tab. 2. Cluster A contained the vegetation types with low ground cover: the plots of Group 1 were sampled in typical garigues of the ultramafic sites of Mediterranean Tuscany, belonging to the Armerio-Alyssetum bertolonii euphorbietosum spinosae (Chiarucci et al. 1995), with a very low shrub cover. This vegetation type is characterised by several endemic plants, the most abundant of which is the nickel hyperaccumulator Alyssum bertolonii (Arrigoni et al. 1983, Chiarucci et al. 1995). Group 2 included similar plant communities, but with a higher shrub cover mainly due to Juniperus oxycedrus subsp. oxycedrus and Phillyrea latifolia. Cluster B contained the plots sampled in xerophilous shrublands. The plots of Group 3 were in open juniper scrub communities, dominated by Juniperus oxycedrus subsp. oxycedrus, in which Cistus salvifolius, Quercus ilex, Phillyrea latifolia and Erica arborea were also abundant. The field layer was dense and rich, being characterised both by garigue species (Centaurea aplolepa subsp. carueliana, Galium corrudifolium, Stachys recta subsp. serpentini, Thymus striatus var. ophioliticus, Trinia glauca, Linum trigynum, Anthericum liliago) and xerophilous perennial graminoids (Bromus erectus, Festuca inops, F. robustifolia, Carex humilis). The plots of Group 4 were sampled in closed juniper scrub communities, in which the broad-leaved evergreen species (Quercus ilex, Arbutus unedo, Phillyrea latifolia) were more abundant than J. oxycedrus subsp. oxycedrus. Cover and richness of the field layer were Vegetation of Tuscan ultramafic soils 117 Table 2. Species frequency (F) according to a five point scale (I, < 20%; II, 20^10%; III, 40-60%; IV, 60-80%; V, 80-100%) and mean cover (C) in the six groups of plots identified by cluster analysis in the Mediterranean data-set. +, mean cover less than 0.01. Species endemic to serpentine or locally exclusive to serpentine are indicated in bold. Cluster/Group A/1 A/2 B/3 B/4 C/5 C/6 F c F c F c F c F c F c Tree layer Erica scoparia I 0.43 Juniperus oxycedrus I 2.14 I 0.42 Fraxinus ornus I 5.71 m 13.08 Arbutus unedo I 0.71 HI 8.75 Phillyrea latifolia I 0.71 III 2.08 Quercus ilex I 3.57 IV 30.42 Erica arborea III 2.83 Quercus pubescens I 1.25 Rhamnus alaternus I 0.83 Hedera helix I 0.08 Quercus suber I 0.83 Smilax aspera I 0.25 Sorbus torminalis I 0.33 Shrub layer Genista januensis II 0.02 lu- 0.15 V 0.36 III 0.23 III 0.17 Cistus incanus II 0.02 ll 0.31 n 1.79 II 0.74 II 1.00 Clematis flammula I 0.03 Osyris alba I 0.06 Cistus salvifolius II 0.10 IV 0.91 IV 7.64 V 1.24 I 0.01 I 0.02 Juniperus oxycedrus V 2.02 V 6.88 V 42.86 V 24.00 V 8.29 III 3.00 Erica arborea II 0.22 II 0.33 V 5.64 V 4.21 V 14.43 III 6.33 Erica scoparia II 0.49 n 1.50 in 4.43 IV 14.57 III 3.86 ni 3.67 Myrtus communis II 0.78 i 0.13 n 4.29 III 2.93 II 7.14 in 2.08 Phillyrea latifolia II 0.48 IV 3.50 V 5.86 V 12.29 V 13.29 IV 5.08 Fraxinus ornus I 0.02 i 0.03 in 1.90 V 3.76 IV 1.01 in 3.92 Quercus ilex I 0.06 Ii 0.44 V 6.86 V 24.00 IV 11.44 V 26.83 Pistacia lentiscus i 0.19 i 0.04 III 2.40 I 0.14 1 0.04 Arbutus unedo IV 0.46 V 8.93 IV 16.00 HI 3.29 Smilax aspera li 0.73 IV 0.24 III 0.27 V 1.87 Lonicera implexa in 0.06 IV 0.21 III 0.14 II 0.13 Asparagus acutifolius li 0.01 I 0.01 III 0.13 III 0.11 Rhamnus alaternus i 0.71 I 0.14 III 5.03 Viburnum tinus li 0.04 II 0.44 I 0.07 ni 1.42 Genista pilosa li 0.04 II 0.11 i 0.03 Quercus pubescens II 0.30. Tamus communis I 0.01 I 0.01 Calicotome villosa I 0.43 Sorbus aria I 0.04 i 0.04 Spartium junceum I 0.03 i 0.03 Ruscus aculeatus I 0.29 III 2.96 in 1.58 Cytisus villosus II 0.09 I 0.04 i 0.02 Lonicera caprifolium I 0.01 I 0.03 i 0.03 Rubus ulmifolius II 0.13 li 0.28 Juniperus communis II 0.57 i 0.13 Rosa sempervirens I 0.07 i 0.04 Sorbus domestica I 0.03 i 0.08 118 A. Chiarucci et al. Cluster/Group A/1 A/2 B/3 F c F c F c Rubia peregrina Pyrus pyraster Field layer Jasione montana ii 0.01 Linum cf. alpinum ii 0.03 Lathmea squamaria ii 0.01 Ceterach officinarum ii 0.01 Anthemis cf. montana i 0.04 Euphorbia nicaeensis subsp. prostrate i 0.28 Festuca arundinacea i 0.01 Polygonum patulum i + Psilurus incurvus i + Sanguisorba minor i 0.01 Scrophularia canina i 0.06 Onosma echioides v 0.62 i 0.25 Sedum album iv 0.11 in 0.18 Anthyllis vulneraria subsp. praepropera iii 0.04 iv 0.16 Silene paradoxa ni 0.03 ii 0.01 Armeria denticulata m 0.06 m 0.26 Allium moschatum ii 0.01 i 0.06 Alyssum montanum ii 0.04 i 0.04 Inula viscosa ii + ii 0.02 Reichardia picroides ii 0.01 i 0.04 Stipa etrusca ii 0.49 ii 0.39 Aim elegans i + i + Medicago minima i + i + Arenaria serpyllifolia i + i + Teucrium montanum v 0.27 v 0.45 iii 0.08 Hemiaria glabra v 0.04 iv 0.03 i 0.01 Iberis umbellata v 0.03 iii 0.02 n 0.01 Linum trigynum iv 0.04 iv 0.06 iv 0.01 Cheilanthes marantae m 0.04 ii + i + Allium sphaerocephalon in 0.01 in 0.01 i + Convolvulus cantabrica in 0.03 i + n 0.01 Hypochoeris achyrophorus ii 0.01 ii 0.01 i + Echium vulgare ii 0.01 i + ii 0.01 Asperula cynanchica v 0.02 in 0.02 Genista januensis in 0.07 Anagallis arvensis ii + Agrostis gigantea i 0.01 Briza media i + Bromus hordeaceus i + Bromus madritensis i + Catapodium rigidum i + Rubus ulmifolius i + Sherardia arvensis i + Trifolium stellatum i + Brachypodium distachyum ii 0.01 Genista pilosa i 0.01 Hieracium pilosella i 0.01 Scabiosa columbaria i + i + B/4 F C C/5 F C i + I 0. C/6 F C I 0.06 I 0.01 01 I + I 0.01 Vegetation of Tuscan ultramafic soils 119 Cluster/Group A/l A/2 B/3 B/4 C/5 C/6 F c F c F c F c F c F c Knautia arvensis i + II 0.05 Prasium majus i + Cerastium ligusticum rv 0.06 III 0.01 I + Dianthus sylvestris subsp. longicaulis v 0.05 v 0.12 II 0.01 I 0.01 Cuscuta epithymum in 0.02 ii + ii + I + Heiichrysum italicum hi 0.84 iii 0.47 iii 0.09 i 0.01 Sedum rupestre v 0.35 IV 0.23 i 0.01 ii 0.04 Iris chamaeiris ii 0.03 iii 0.09 i 0.01 II 0.02 Anthericum liliago v 0.03 IV 0.03 iii 0.01 HI 0.01 i + Plantago holosteum rv 0.80 iii 0.30 II 0.03 i 0.03 i 0.57 Stachys recta subsp. serpentini rv 0.11 III 0.09 v 0.12 I 0.01 i 0.01 Alyssum bertolonii V 0.58 v 0.36 II 0.02 i 0.01 II 0.02 Centaurea aplolepa subsp. carueliana v 1.38 v 1.11 IV 0.11 III 0.04 II 0.24 Thymus striatus var. ophioliticus V 0.63 v 0.64 III 0.14 I 0.01 III 0.15 Festuca robustifolia V 0.46 III 0.43 v 1.21 iv 0.74 III 0.12 Galium corrudifolium rv 0.14 v 0.21 V 0.55 iv 0.11 III 0.05 Trinia glauca subsp. glauca rv 0.15 IV 0.17 v 0.08 v 0.27 in 0.03 Koeleria splendens rn 0.19 iii 0.11 i 0.01 I 0.04 in 0.09 Euphorbia spinosa i 0.22 iii 1.88 ii 1.50 ii 0.08 i 0.36 Linum tenuifolium r 0.01 I 0.01 i 0.01 i + i + Centaurium pulchellum i 0.06 I + ii 0.06 i + i 0.01 Bromus erectus rv 0.89 iv 1.61 v 14.43 v 7.07 IV 1.31 Potentilla hirta li 0.17 v 0.19 ii 0.02 ii 0.03 in 0.02 Vincetoxicum hirundinaria i + i 0.01 II 0.02 i + in 0.02 Cistus salvifolius li 0.36 i 0.01 Biscutella pichiana subsp. pichiana li 0.01 i + Sesleria cf. tenuifolia li 1.00 i 5.71 Centaurium erythraea i + i + Dorycnium hirsutum li 0.06 ii 0.01 IV 0.04 li 0.01 Dactylis hispanica i 0.02 II 0.16 II 0.43 li 0.12 Danthonia alpina Ii 0.12 II 1.43 i 2.14 Cytisus decumbens i 0.01 i + i 0.01 Fumana procumbens i 0.05 i 0.01 Cytinus hypocistis i 0.01 ii 0.01 I + i 0.01 Staehelina dubia i 0.13 i 0.01 i 0.01 Teucrium polium i 0.01 i 0.03 Hieracium piloselloides I 0.01 II 0.02 i 0.01 Brachypodium sylvaticum i 0.01 li 0.29 Centaurea nigra i 0.01 i + Melittis melissophyllum i 0.01 Polygala flavescens i 0.03 i 0.01 i 0.01 Carex hallerana I 0.21 i 0.43 i 0.07 Leontodon villarsii i + Festuca inops rv 1.92 v 3.91 IV 1.47 III 0.21 li 0.29 I + Carex humilis li 0.44 ii 0.14 III 4.47 ii 0.29 m 0.36 i 0.08 Stachys officinalis r + ii 0.05 in 0.05 iii 0.03 Asparagus acutifolius i + i 0.01 li 0.03 II 0.02 Rubia peregrina ii 0.04 v 0.29 IV 0.29 v 0.88 Brachypodium rupestre i 0.21 iii 4.57 IV 3.96 iii 0.36 Cyclamen repandum i 0.01 II 0.79 in 0.15 rv 0.36 Tanacetum corymbosum ii 0.07 II 0.09 in 0.08 i + Filipendula vulgaris i 0.01 i 0.04 i + Serratula tinctoria I + i 0.01 i + 120 A. Chiarucci et al. Cluster/Group A/1 A/2 B/3 B/4 C/5 C/6 FC FC FC FC FC FC Carex flacca . . in 0.30 III 0.59 I + Viola alba subsp. dehnhardtii . . i + III 0.02 II 0 Asplenium onopteris I. serpentini . . i + I 0 Viburnum tinus . . i 0.01 I + Tamus communis I 0.01 II 0 Bromus ramosus I + I 0 Peucedanum cervaria I + I + Hedera helix II 0 Melica uniflora I + Ruscus aculeatus I 0 Carex distachya I 0 Cephalanthera rubra I + Cerastium arvense I + Silene nutans I + Smilax aspera I + Table 3. Soil and environmental data (mean ± s.e.) for the six groups identified by cluster analysis in the Mediterranean data-set. When analyses were applied to log-transformed data the standard error range is given in brackets. Different letters indicate significant differences between groups of plots. Duncan multiple-range test at P < 0.05. Cluster/Group n A/1 9 A/2 8 B/3 7 B/4 7 C/5 7 C/6 12 Rockiness (%) 26.4 + 7.8 18.5 ±5.8 9.4 ± 1.9 4.3 ± 1.1 8.0 ±4.7 1.1 ±0.5 Stoniness (%) 51.7 + 6.8 56.3 ±7.8 17.0±5.5 4.9 ± 2.6 6.6 ±4.1 1.3 ±0.4 Insolation 2139 ± 131 ab 2230±121 a 2318 + 88 a 2247±143 a 1933 ± 143 ab 1833 ± 100 b LOI 10.6 a (9.7-1.5) 10.2 a (9.5-0.9) 18.4 b (16.0-18.1) 17.0 b (15.8-21.5) 20.3 b (18.3-22.5) 21.0 b (18.7-23.6) pH 7.4 + 0.1 a 7.2 + 0.1 ab 6.9 ±0.1 c 6.9 ±0.1 be 6.7 ±0.1 c 6.7 + 0.1 c Ca 444 a (340-579) 577 ac (439-758) 899 ab (614-1318) 1028 ab (877-1205) 1651 b (1140-2392) 1332 be (1105-1605) Mg 1253 ab (983-1597) 799 a (687-928) 1696 ab (1192-2414) 1506 ab (1296-1750) 2598 b (1700-3969) 1768 ab (1516-2062) Mg/Ca 2.9 ±0.3 a 1.9 ± 0.4 b 1.9±0.1 b 1.8 ±0.4 b 1.7 ±0.2 b 1.6 ± 0.2 b Cu 0.2 (0.2-0.3) 0.2 (0.2-0.3) 0.3 (0.3-0.3) 0.4 (0.3-0.4) 0.3 (0.3-0.4) 0.5 (0.3-0.6) Fe 2.2 (2.0-2.5) 2.3 (2.1-2.6) 2.8 (2.2-3.4) 2.7 (2.5-3.0) 3.2 (2.7-3.7) 3.4 (3.0-3.7) K 45 a (32-63) 39 a (32-46) 106 ab (69-162) lOOab (83-122) 136 b (82-225) 171 b (127-229) Mn 3.1 a (2.5-3.7) 2.7 a (1.9-3.7) 8.5 b (7.0-11.3) 19.8 c (17.0-23.0) 7.9 b (5.8-10.7) 10.0 be (7.3-13.6) Ni 3.4 a (2.8-4.0) 3.6 a (2.8-4.6) 8.2 b (7.4-9.1) 6.4 b (7.6-7.0) 8.8 b (7.9-9.8) 6.6 b (5.6-7.8) Vegetation of Tuscan ultramafic soils 121 q + Altitude lope1' Insolation3 2 3 2 Stoniness Fig. 2. CCA ordination diagram of the environmental variables and 50 plots of the Mediterranean data-set. The numbers refer to the group of plots (see text). The environmental variables Mg, Ca, Mn, Cu and Fe are not shown, because of their short length. reduced because of the higher shrub cover. The garigue species and the xerophilous graminoids were almost totally absent, being partly replaced by Brachypodium rupestre and Rubia peregrina. The vegetation of these two groups has not yet been described from a phytosociological point of view. Cluster C, linked to cluster B, contained closed woody communities. Group 5 combined the plots sampled in closed maquis a few metres high, locally called "forteto", in which the Mediterranean broad-leaved evergreen species, together with Erica scoparia and E. arborea, played the main structural role, whereas /. oxycedrus subsp. oxycedrus was scarce. The field layer consisted of Brachypodium rupestre and several woodland species (Rubia peregrina, Stachys officinalis, Cyclamen repandum, Viola alba subsp. dehnhardtii). This vegetation type has been referred to the Viburno-Quercetum ilicis ericetosum (marchiori & Tornadore Marchiori 1977, Chiarucci 1994). The plots of Group 6 were in the most evolved communities, similar to Group 5 but with a structure often differentiated into three layers. This vegetation type was assigned to the Viburno-Quercetum ilicis ornetosum (Marchiori 122 A. Chiarucci et al. & tornadore marchiori 1977, chiarucci 1994). The field layer had a very low cover and low species richness. Soil and physical environment The main soil features are reported in Tab. 3. The results show that organic matter content and availability of metals increased from garigue to woodland vegetation types, whereas the pH, rockiness and stoniness decreased. Insolation was higher in the shrubland community than in maquis and woodlands. In the garigue it had intermediate values. Chromium, which usually has high total concentrations in ultramafic soils (Brooks 1987), was below the detection limit (< 0.5 |J.g/g) in the soil extracts. The Mg/Ca quotient, which has been regarded as important in controlling some serpentine floras (Proctor & Nagy 1992), was significantly higher only under the garigue vegetation. Gradient analysis The CCA ordering of plots and environmental variables is shown in Fig. 2. The cumulative percentage of variance explained by the first two axes accounted for 25.4% (19.9% and 5.5%) of the species data and 56.8% (44.6% and 12.2%) of the species-environment relations; the eigenvalue of the first canonical axis was highly significant (P < 0.01). The plot scattering according to the first two axes (Fig. 2) showed an evolutive trend from garigues to garigues with shrubs, open juniper scrub, closed juniper scrub, maquis and woods, from the positive to the negative values on the first axis. The first axis was positively correlated with rockiness, stoniness and pH and negatively with loss on ignition, whereas the second axis was positively correlated with altitude (Tab. 4). Exchangeable cations were negatively, but not significantly, correlated with the first axis, indicating increasing availability of these elements in the soil from the less to the more evolved communities. Inland data set Species composition and richness The two main clusters (not shown) pertained to the plant communities with high (A) and low (B) degrees of bare soil, differing in structure and species diversity (Tab. 5). The two groups of cluster A corresponded to Groups 1 and 2 of the Mediterranean data set but were slightly different in terms of flora (Tab. 6). In fact, the plots of Group 1 belonged to the Armerio-Alyssetum bertolonii typicum, which is reported for all the ultramafic sites of inland Tuscany (Chiarucci et al. 1995). Group 2 pertained to similar vegetation but with a higher ground cover and with a higher shrub cover, Juniperus oxycedrus subsp. oxycedrus and Fraxinus ornus being the most abundant shrubs. Table 4. Intraset correlations of environmental variables with the first two axes of CCA ordination in the Mediterranean data-set. Statistically significant values are indicated in bold. Correlation coefficients Variable Axis 1 Axis 2 Altitude 0.063 0.515 Slope 0.353 -0.263 Insolation 0.263 -0.263 Rockiness 0.780 0.156 Stoniness 0.900 -0.049 LOI -0.648 -0.056 pH 0.605 0.343 Ca -0.380 -0.091 Cu -0.268 -0.020 Fe -0.362 0.018 K -0.429 -0.348 Mg -0.273 -0.168 Mn -0.454 -0.047 Na -0.276 -0.250 Ni -0.399 -0.260 Vegetation of Tuscan ultramafic soils 123 Table 5. Structural features and species diversity data (mean ± s.e.) of plots belonging to the six groups identified by cluster analysis in the inland data-set. Cluster/Group A/1 A/2 B/3 B/4 B/5 n 7 6 7 6 4 Tree layer cover (%) 0.7 ± 0.7 _ 22.9 + 5.9 14.2 ±3.0 75.0 + 2.0 Shrub layer cover (%) 2.8 ±1.1 35.9 ±12.3 45.0 + 13.7 13.8 ±11.3 32.5 ± li.i Field layer cover (%) 30.7 ± 6.4 49.5 ± 10.1 85.0 ±2.7 75.8 ±6.0 14.3 ± 6.2 Total species richness 24.7 ± 1.8 28.3 ± 2.0 21.6 ±0.9 25.7 ±2.3 20.8 ± 0.9 Tree layer richness 0.1 ±0.1 0.0 ±0.0 0.9 ±0.1 1.0 ±0.0 2.8 ± 0.8 Shrub layer richness 1.0 ± .4 4.3 ±0.6 5.4 ±1.0 3.0 ±0.7 8.8 ± 1.1 Field layer richness 23.6 ± 2.0 24.0 ±1.9 15.7 ±1.1 22.2 + 1.7 11.5 ± 1.2 Table 6. Species frequency (F) according to a five point scale (i, < 20%; ii, 20-40%; iii, 40-60%; iv, 60-80%; v, 80-100%) and mean cover (C) in the five groups of plots identified by cluster analysis in the inland data-set. + mean cover less than 0.01. Species endemic to serpentine or locally exclusive are indicated in bold. Cluster/Group A/1 A/2 b/3 b/4 b/5 F C F C F c F C F C Tree layer Pinus pinaster I 0.71 v 20.71 v 14.17 Fraxinus ornus v 6.75 Quercus pubescens v 55.75 Ostrya carpinifolia ii 12.50 Quercus cerris ii 3.75 Sorbus aria ii 2.50 Shrub layer Erica scoparia i 0.50 i 0.13 iii 17.86 ii 11.80 Erica arborea i 0.07 iv 26.43 iii 0.25 Rosa agrestis n 0.17 Juniperus oxycedrus subsp. oxycedrus m 2.29 v 34.17 i 0.09 i 0.03 v 2.38 Fraxinus omus I + v 2.45 ii 0.13 v 10.13 Sorbus aria i 0.50 ii 0.50 Quercus pubescens n 0.17 ni 0.75 Genista januensis v 0.30 rv 0.17 ii 0.12 ii 0.05 Spartium junceum ii 0.13 ii 0.43 i 0.05 ii 0.01 Rubus ulmifolius i 0.08 rv 0.16 v 8.33 Cistus salvifolius in 0.13 Inula viscosa i 0.07 Rhamnus alaternus ii 0.29 ii 0.08 Asparagus acutifolius ii 0.13 i 0.02 Pinus pinaster in 1.76 iii 1.58 Juniperus communis i 0.09 iii 0.08 ii 0.50 Ruscus aculeatus i 0.03 ii 0.75 Prunus spinosa iv 8.28 Cornus mas m 0.33 Prunus avium hi 0.23 Rosa arvensis in 0.30 Clematis vitalba ii 0.25 Coronilla emerus ii 0.50 Cytisus sessilifolius ii 0.38 Lonicera caprifolium ii 0.63 124 A. Chiarucci et al. Cluster/Group Ostrya carpinifolia Sorbus domestica Ulmus minor Field layer Cerastium ligusticum Centaurium pulchellum Filago germanica Arenaria serpyllifolia Asterolinon linum-stellatum Fumana procumbens Jasione montana Stipa tirsa Vulpia ciliata Aira eiegans Anthyllis vulneraria subsp. praepropera Armeria denticulata Sedum album Echium vulgare Hypochoeris achyrophorus Minuartia laricifolia subsp. ophiolithica Plantago holosteum Centaurea cf. jacea Hypericum perforatum Asplenium cuneifolium Aira caryophyllea Asplenium adiantum-nigrum van serpentini A/1 F C A/2 F C B/3 F C B/4 F C ni 0.04 II 0.01 II + I 0.01 I + I 0.07 I + I + I + IV 0.01 rv o.i5 IV 0.46 m 0.06 ni + i + I o.oi HI 0.77 I 0.01 I + I 0.01 I + I + ni o.oi i + i o.oi V 0.37 IV 0.06 ni o.oi II + I + I 0.01 Centaurium erythraea I + Ceterach officinarum I + Geranium robertianum I + Polygala nicaeensis I + Anthericum liliago n 0.01 Inula viscosa n 0.29 Plantago lanceolata n 0.01 Leucanthemum pachyphyllum i 0.21 Gymnadenia cf. conopsea i + Quercus ilex i + Hieracium pilosella rv 0.01 Euphorbia nicaeensis subsp. prostrata II 0.02 Linum tenuifolium I + Cistus incanus I 0.17 Silene paradoxa III 0.07 IV 0.07 + Dianthus sylvestris subsp. longicaulis V 0.05 IV 0.10 I 0.01 Herniaria glabra V 0.09 I + Artemisia alba IV 1.29 V 1.40 II 0.67 Sedum rupestre V 0.59 ni 0.03 II 0.01 Cheilanthes marantae III 0.19 i 0.02 I + Cistus salvifolius II 0.51 i 0.83 I + Linum trigynum V 0.02 V 0.01 rv 0.01 Stipa etrusca III 12.24 V 1.63 i 0.03 i 0.02 Alyssum bertolonii V 0.34 V 0.43 i 0.03 ii 0.01 Cuscuta epithymum III 0.01 III 0.01 i + i + Helichrysum italicum IV 0.56 V 1.43 m 0.25 v 0.28 Stachys recta subsp. serpentini III 0.01 ni 0.15 in 0.04 II 0.01 B/5 F C II 0.50 II 0.20 II 0.10 Vegetation of Tuscan ultramafic soils 125 Cluster/Group A/1 A/2 B/3 B/4 B/5 F c F c F C F c F c Sanguisorba minor subsp. muricata iv 0.47 v 0.19 iii 0.04 ii 0.08 Trinia glauca subsp. glauca iii 0.03 v 0.04 ii 0.06 iii 0.04 Thymus striatus var. ophioliticus v 2.64 v 2.03 ii 0.02 v 0.77 Reichardia picroides iii 0.01 ii 0.02 iii 0.01 iii 0.02 Danthonia alpina iii 0.45 ii 9.23 iii 1.93 v 31.00 Potentilla hirta iii 0.30 i 0.01 iv 0.03 iv 0.04 Hieracium piloselloides ii 0.01 i + iii 0.05 iv 0.02 Convolvulus cantabrica i 0.01 i + i 0.02 Genista januensis ii 0.08 i 0.11 iv 0.13 Aster linosyris i + i 0.01 ii 0.02 Centaurea aplolepa subsp. carueliana in 0.93 iii 0.03 v 0.69 Allium sphaerocephalon m 0.02 iii 0.01 Scorzonera austriaca i 0.01 iii 0.03 iv 0.12 Centaurea rupestris subsp. rupeslris i 0.03 iii 0.54 v 3.13 Knautia arvensis iv 0.50 iii 0.34 i 0.13 Hippocrepis comosa iv 0.06 i 0.01 iii 0.05 Lotus comiculatus i + i + Carlina corymbosa in 0.03 v 0.11 iv 0.11 Briza media i 0.01 i + Holoschoenus australis i 7.14 i 0.17 Festuca inops v 7.19 v 1.63 iii 0.21 iv 2.08 ii 0.03 Bromus erectus in 0.79 v 27.80 iii 8.00 v 28.67 ii 0.03 Galium corrudifolium in 0.04 v 0.12 v 0.12 v 0.07 m 0.26 Festuca robustifolia ii 1.46 i + rv 5.10 v 9.08 ii 0.01 Carex humilis i 0.36 i 0.33 in 0.19 iv 0.50 ii 3.00 Vincetoxicum hirundinaria i + i 0.03 rv 0.07 v 0.28 ii + Melica ciliata i 0.03 iii 0.38 ii + Dorycnium hirsutum n 0.01 iv 0.25 ii 0.03 Silene vulgaris i + m 0.03 Quercus pubescens i + ii + ii + ii 0.01 Crepis leontodontoides i + ii 0.03 Phleum bertolonii i + ii 0.01 Viola alba subsp. dehnhardtii i + iv 0.40 Carex flacca n 0.13 i 0.21 iii 0.63 Brachypodium rupestre i 0.83 v 62.14 ii 1.00 v 8.63 Filipendula vulgaris i 0.01 ii 0.01 Stachys officinalis i 0.01 ii 0.50 Tanacetum corymbosum i 0.01 in 0.02 Agrostis gigantea ii 0.12 ii 0.03 Clinopodium vulgare iv 0.03 Carex divulsa m 0.28 Cephalanthera rubra in 0.01 Arabis hirsuta ii 0.01 Cruciata glabra ii 0.50 Dactylis hispanica ii 0.03 Dorycnium pentaphyllum ii 0.03 Festuca heterophylla ii 0.38 Hedera helix ii 0.05 Inula conyza ii 0.01 Lathyrus latifolius ii 0.08 Lonicera caprifolium ii 0.05 Peucedanum cervaria ii 0.02 Torilis arvensis ii 0.01 126 A. Chiarucci et al. Table 7. Soil and environmental data (mean ± s.e.) for the six groups identified by cluster analysis in the inland data-set. When analyses were applied to log-transformed data the standard error range is given in brackets. Different letters indicate significant differences between groups of plots. Duncan multiple-range test at P < 0.05. Cluster/Group n A/1 7 A/2 6 B/3 7 B/4 6 B/5 4 Rockiness (%) 14.0 + 6.8 8.5 + 2.7 5.0 ± 1.2 9.2 ± 2.9 0.3 ±0.3 Stoniness (%) 42.1 ±8.0 16.7 ±8.2 1.3 ±0.3 4.3 ± 1.8 25.0±21.7 Insolation 2140 ± 110 ab 2344 ± 56 a 1907 +153 b 2235 ± 50 ab 2031±158 b LOI 10.8 a (9.6-12.0) 13.5 a (12.4-14.7) 21.3 b (20.3-22.4) 18.6 b (17.6-19.6) 21.3 b (20.7-21.9) pH 7.1 ±0.1 a 6.9 ±0.1 a 6.5 ±0.1 b 6.5 ±0.1 b 6.7 ±0.1 b Ca 667 a (421-1058) 631 a (523-762) 827 a (744-919) 1611 a (1014-2560) 1668 a (1222-2276) Mg 1460 ab (931-2288) 1170 a (1035-1323) 1651 ab (1422-1918) 4193 b (2624-6699) 1620 ab (1328-1976) Mg/Ca 2.2 ±0.1 ab 1.9 ±0.2 be 2.0 ±0.2 a 2.7 ±0.3 ab 1.3±0.5c Cu 0.2 a (0.2-0.3) 0.3 a (0.2-0.3) 0.3 ab (0.3-0.4) 0.3 ab (0.3-0.4) 0.4 b (0.4-0.5) Fe 1.7 a (1.6-1.8) 2.9 b (2.4-3.4) 3.0 b (2.8-3.3) 2.4 ab (2.0-3.0) 3.5 b (3.0^.2) K 76 a (44-132) 49 a (40-61) 62 a (56-69) 175 a (100-306) 169 a (133-214) Mn 2.1 a (1.6-2.6) 4.4 b (3.2-5.9) 7.1 b (5.7-8.9) 7.2 ab (5.0-10.3) 10.8 b (6.3-18.1) Ni 2.6 a (2.3-2.9) 4.5 ab (3.5-5.7) 10.4 c (8.6-12.5) 13.7 c (11.6-16.1) 7.5 ab (5.1-10.9) Cluster B contained markedly different vegetation types. Group 3 consisted of plots in grassland dominated by Brachypodium rupestre under a relatively dense pine (Pinus pinaster) canopy (20% of cover on average). The strong dominance of Brachypodium rupestre (mean cover 62%) reduced species richness in the field layer. Group 4 was formed by less strongly dominated grasslands under a lower pine cover (14% of cover on average). These evolve from garigues similar to those of group 1 because of the pine canopy, which protects against erosion and is responsible for an increase in nutrient input (Chiarucci & De Dominicis 1995, Chiarucci 1996, Chiarucci et al. 1998) and probably they also evolve into the grasslands of Group 3. The plots of Groups 3 and 4 were sampled exclusively on Mt. Ferrato. The plots of Group 5, sampled in the Upper Tiber Valley, were recorded in the most evolved vegetation, i.e. woodlands dominated by Quercus pubescens and Fraxinus ornus. These communities were not common and, as observed by Pichi Sermolli (1948), were almost exclusively associated with less xeric conditions, such as slope-bottoms, or close to the ultramafic/non-ultramafic lithological ecotone. Vegetation of Tuscan ultramafic soils 127 o T— + o Fig. 3. CCA ordination diagram of the environmental variables and 30 plots of the inland data-set. The numbers refer to the group of plots (see text). The environmental variables Ca, K, Mg and Na are not shown, because of their short length. Soil and physical environment The main soil features of the vegetation groups of inland areas (Tab. 7) showed a picture similar to that reported for the Mediterranean areas. The plot groups in the inland data-set did not represent a natural succession, as those of the Mediterranean data-set, but included anthropogenically induced vegetation, such as the pine-dominated communities. Despite all this, pH, organic matter content, and metal availability were correlated with vegetation cover and structure, implying that the soil under garigue and scrub vegetation may be transformed by the growth of forests, such as invading pine trees. Gradient analysis The CCA ordering of plots and environmental variables is shown in Fig. 3. The cumulative percentage of variance accounted by the first two axes was 35.4% (19.1% and 16.3%) of the species data and 48.9% (26.3% and 22.6%) of the species-environment relationships (Fig. 3). The eigenvalue of the first canonical axis was highly significant (P < 0.01). The first axis, 128 A. Chiarucci et al. Correlation coefficients Variable Axis 1 Axis 2 Altitude 0.464 0.697 Slope 0.574 0.361 Insolation -0.095 0.351 Rockiness -0.673 0.077 Stoniness -0.047 0.746 LOI 0.428 -0.743 pH -0.167 0.775 Ca - - Cu 0.300 -0.343 e 0.412 -0.245 K - - Mg - - Mn 0.392 -0.495 Na - - Ni 0.028 -0.742 Table 8. Intraset correlations of environmental which differentiated the woods of Group 5 with variables with the first two axes of CCA reSpect to other vegetation groups, was correlated ordination in the inland data-set. Statistically ... , ... , , .. , . .,. . ... . .. , ,, 3 positively with slope and negatively with significant correlation are marked in bold. r J c ~. ■ rockiness. The second axis, which differentiated (a) Groups 3 and 4, (b) Group 5 and (c) Groups 1 and 2, was positively correlated with altitude, stoniness and pH, and negatively with organic matter, Mn and Ni content of the soil (Tab. 8). DISCUSSION The results indicate that the exchangeable fraction of soil metals was higher under the more evolved and structured communities, in both natural and anthropogenic environments, suggesting that soil metal content is not the most important limiting factor for the vegetation of Tuscan ultramafic soils. The soluble fraction of chromium was, in any case, too small to affect the _ vegetation (Brooks 1987, Pandolfini & Pancaro 1992, Robinson et al. 1996). The increased availability of metals in the soil under the more evolved vegetation types may be a direct consequence of the lower pH. Robinson et al. (1996) showed that metal solubility in New Zealand serpentines increased exponentially as the pH of the extracting solution was lowered. The effect of closed woody vegetation on soil may be to lower the pH by humic decay under moist conditions and therefore induce a higher availability of metals. The hypothesis that metal content is not the most important limiting factor causing the infertility of Tuscan serpentine soils was recently advanced in pedological (Angelone et al. 1991, 1993) and vegetation (Chiarucci & De Dominicis 1995, Chiarucci 1996, Chiarucci et al. 1998) studies. The results of these studies were at variance with previous reports, and these authors emphasised the importance of the metal fraction available to plants as opposed to total metal concentrations. Many other authors claim that soil metal toxicity is not the most important limiting factor for serpentine vegetation. Carter et al. (1987) did not find any evidence that nickel was the cause of the serpentine infertility in the keen of Hamar, Shetland Islands. Kruckeberg (1992) did not find any evidence that cobalt, chromium, iron and nickel affect plant growth in the ultramafic soils of western North America. In New Zealand, Lee (1992) observed that only in some southern ultramafics is nickel likely to reduce plant growth. In a review, Proctor & Nagy (1992) stated that many assumptions about causal roles for nickel in causing the unusual serpentine vegetation are unfounded and in-depth studies have often disproved the importance of this element. One of the most important factors controlling the vegetation of Tuscan ultramafic soils appears to be drought stress due to topographical position. In both Mediterranean and inland sites, the insolation of juniper scrub communities, a relatively undisturbed vegetation type where the serpentine endemics are found, was significantly higher than that of the sites with a woodland vegetation cover. The index used in the present survey indicates the hours per year of overhead sunlight to give the radiation the site actually receives in one year, given Vegetation of Tuscan ultramafic soils 129 its aspect, slope, and latitude, but the differences between sites might also be more important if considering summer drought. Water stress, together with soil nutritional deficiencies, constantly limits vegetation development. Periodic drought and other exceptional stresses may kill off the forest species, interrupting vegetation dynamics and thereby causing reversion to stunted and open plant communities and promoting soil erosion. For the serpentine soils of the Keen of Hamar, Shetland Islands, where rainfall is abundant throughout the year. Carter et al. (1987) suggested that summer droughts of seven or more consecutive days, which happen on average once every 3-5 years, may exhaust the water reserves of the soil. This leads to the development of xerophilous, mainly annual, plant communities even in the humid oceanic climatic zone. There is now also strong evidence that nutrient availability is another important factor limiting plant growth and hence vegetation development in Tuscan ultramafic soils. The lower pH under woody plant communities not only increases the availability of plant nutrients such as K, Zn, Cu, Fe, but also facilitates their uptake. The N and P available (not measured) under maquis and woodland vegetation types are greater because of the higher organic matter content. Some recent studies showed that the addition of small amounts of nitrogen, phosphorus and potassium to garigue vegetation induced a strong increase in ground cover and biomass (Chiarucci et al., unpubl. data). Higher nutrient availability has also been reported to alleviate the effects of drought (Grime & Curtis 1976). Proctor & Nagy (1992) considered the low nutrient content of the soil a "key feature" in causing serpentine infertility in many areas. On the other hand, Ca-deficiency or the Mg/Ca ratio, considered very important limiting factors for serpentine vegetation (Brooks 1987, Baker et al. 1992, Roberts & Proctor 1992), did not appear as very important in these ultramafic soils. In fact the addition of calcium, alone or with nitrogen, phosphorus and potassium (NPK), led to a non-significant increase of cover and biomass with respect to controls and NPK alone (Chiarucci et al. unpubl. data). This was probably due to the fact that in these soils the Mg/Ca quotient is relatively low, averaging 1.5 to 3.0. The development of more-structured woody communities positively improves soil structure and fertility, allowing a better plant growth by positive feedback. The natural dynamics towards more evolved vegetation types lead to the disappearance of the endemic species, linked to the less evolved soils, and the spread of the more competitive forest species. This implies that the persistence of garigue and open scrub vegetation, in which the serpentine endemics grow (Pichi Sermolli 1948, Pignatti 1982, Arrigoni et al. 1983, Chiarucci et al. 1995), might require periodic disturbance or a factor that limits vegetation dynamics. Although specific data are not available, it can be inferred that forest fires are not rare, especially in the Mediterranean sites, but normally do not result in reversion to garigue vegetation. Likewise, anthropogenic disturbance over the last 3000-5000 years would have been insufficient for the evolution of serpentine endemic species such as the endemic nickel hyperaccumulator Alyssum bertolonii. Unlike the open juniper scrubs, the garigues of the Armerio-Alyssetum bertolonii association are not always located in areas with less favourable ecological conditions, such as xeric slopes. In addition, soils under the garigue plant communities show the lowest concentrations of potentially toxic metals. The lack of correlation between environmental limiting factors and garigue vegetation suggests that the spread of garigues has been partially favoured by human activity through destruction of more evolved vegetation types, such as the juniper scrub communities and the evergreen maquis. 130 A. Chiarucci et al. According to this picture, the Juniperus oxycedrus subsp. oxycedrus scrub communities are the most natural vegetation type on xeric slopes and their successional dynamics, towards closed and structured woodland communities, is prevented by nutrient deficiency and periodic reversion due to special stress factors such as drought. As is known for other environments (Grime 1990, van der Maarel 1996), these droughts may be the driving force for the serpentine vegetation of Tuscany, creating gaps for new species to occupy. Xerophilous and heliophilous plants, such as the serpentine endemic Alyssum bertolonii, find optimal ecological conditions in open patches between shrubs. During evolutionary time, soil metals might have induced a selective pressure on plant populations growing in these patches, where the low-nutrient content of the soil and the drought stress cannot alleviate metal toxicity, allowing the selection of specialised taxa. True forest communities, on the other hand, can only be found in more favourable topographic conditions, where water stress is lower, exceptional droughts are less frequent and soil can accumulate. Long-term monitoring of permanent plots can test this hypothesis. Acknowledgement: The authors acknowledge Dr. Bruno Foggi, Dr. Ettore Putorti, Mr. Vincenzo Gonnelli and Mr. Davide Morrocchi for help during field sampling in the different areas. They also thank Dr. Ermanno Busoni for providing data on the index of normalised insolation, Dr. Claudia Angiolini, Dr. Simona Maccherini, and three anonymous referees for critical observations on the text. REFERENCES ANGELONE M., VASELLIO., BlNI C. & CORADOSSI N. (1993): Pedogeochemical evolution and trace elements availability to plants in ophiolitic soils. Sci. Total Environm. 129: 291-309. ANGELONE M., VASELLI O., BlNI C, CORADOSSI N. & PANCANI M.G. (1991): Total and EDTA-extractable element contents in ophiolitic soils from Tuscany (Italy). Z. Pflanzenerna.hr. Bodenk. 154: 217-223. ARRIGONI P.V. (1980): Aspetti corologici della flora sarda (Chorological aspects of the Sardinian flora). Lav. Soc. ltd. Biogeogr. 8: 83-109. ARRIGONI P. V., RlCCERlC. & MAZZANTI A. (1983): La vegetazione serpentinicola del Monte Ferrato di Prato in Toscana (The serpentinicolous vegetation of the Monte Ferrato in Tuscany). Centre di Scienze Naturali, Prato. BAKER A.J.M., PROCTOR J. & REEVES R.D. (eds.) (1992): The vegetation ofultramafic (serpentine) soils. Intercept, Andover. BARTORELLIU. (1967): Tavole numeriche dell'assolazione annua (Numerical tables of annual insolation). Ann. Accad. ltd. Sci. Forest. 16: 61-83. BROOKS R.R. (1987): Serpentine and its vegetation. A mdtidisciplinary approach. Dioscorides, Portland. CARTER S.P., PROCTOR J. & SLINGSBY D.R. (1987): Soil and vegetation of the Keen of Hamar serpentine, Shetland. J. Ecol. 75: 21-42. CHIARUCCI A. (1994): Successional pathway of Mediterranean ultramafic vegetation in central Italy. Acta Bot. Croat. 53: 83-94. CHIARUCCI A. (1996): Species diversity in plant communities on ultramafic soils in relation to pine afforestation. J. Veg. Sci. 7: 57-62. CHIARUCCI A. & De DOMINICIS V. (1995): Effects of pine plantations on ultramafic vegetation of central Italy. Israel J. PI. Sci. 43: 7-20. CHIARUCCI A., FOGGI B. & SELVI F. (1995): Garigue plant communities of ultramafic outcrops of Tuscany (Italy). Webbia 49: 179-192. CHIARUCCI A., RlCCUCCI M., CELESTIC. & De DOMINICIS V. (1998): Vegetation-environment relationships in the ultramafic area of Monte Ferrato, Italy. Israel J. PL Sci. 46: (in press). GRIME J.P (1990): Mechanisms promoting floristic diversity in calcareous grasslands. In: HlLLIER S.H., WALTON D.W.H. & WELLS D.A. (eds.), Calcareous grasslands: ecology and management, Bluntisham Books, Bluntisham, pp. 51-56. Vegetation of Tuscan ultramafic soils 131 grime].?. & curtis A.V. (1976): The interaction of drought and mineral nutrient stress in calcareous grasslands. J. Ecol. 64: 975-998. kruckeberg A.R. (1992): Plant life of western North American ultramafics. In: roberts B.A. & proctor I. (eds.), The ecology of areas with serpentinized rocks, A world view, Kluwer, Dordrecht, pp. 31-74. lee W.G. (1992): The serpentinized areas of New Zealand, their structure and ecology. In: roberts B.A. & proctor I. (eds.). The ecology of areas with serpentinized rocks, A world view, Kluwer, Dordrecht, pp. 375-417. marchiori S. & tornadore marchiori N. (1977): Lineamenti vegetazionali del Monte Pelato -Castiglioncello (Livorno) (Vegetational aspects of the Monte Pelato - Castiglioncello (Livorno)). Atti Soc. Tose. Sei. Nat. Pisa Mem., Ser. B. 84: 7-15. messeri A. (1936): Ricerche sulla vegetazione dei dintorni di Firenze. IV. La vegetazionedelle rocce ofiolitiche del Monte Ferrato (presso Prato) (Research on the vegetation of the surroundings of Florence. IV. The vegetation of the ophiolitic rocks of Monte Ferrato (near Prato)). Nuovo Giorn. Bot. Ital. NS. 43: 277-372. moraldo B. (1986): II genere Stipa (l.) in Italia (The genus Stipa in Italy). Webbia 40: 203-278. pandolfini T. & Pancaro L. (1992): Biogeochemical survey of some ophiolitic outcrops in Tuscany. Flora 187: 341-351. pichi Sermolli R. (1948): Flora e vegetazione delle serpentine e delle altre ofioliti dell'Alta valle del Tevere (Toscana) (Flora and vegetation of the serpentinites and other ophiolites of the Upper Tiber Valley (Tuscany)). Webbia 6: 1-380. PIGNATT1 S. (1982): Flora d'ltalia (Flora of Italy). 1-3. Edagricole, Bologna. podani J. (1993): SYN-TAX version 5.0. User's guide. Scientia, Budapest. podaní I. (1994): Multivariate data analysis in ecology and systematics. A methodological guide to the SYN-TAX 5.0 package. SPB Academic Publ., The Hague. proctor I. & nagy L. (1992): Ultramafic rocks and their vegetation: an overview. In: baker A.I.M., proctor I. & Reeves R.D. (eds.), The vegetation of ultramafic (serpentine) soils, Intercept, Andover, pp. 469-494. roberts B.A. & proctor I. (1992): The ecology of areas with serpentinized rocks. A world view. Kluwer, Dordrecht. Robinson B.H., Brooks R.R., Kirkman J.H., Gregg P.E.H. & Gremigni P. (1996): Plant-available elements in soils and their influence on the vegetation over ultramafic ("serpentine") rocks in New Zealand. J. Roy. Soc. New Zealand 26: 455-^66. sasse F. (1979a): Untersuchungen an Serpentinstandorten in Frankreich, Italien, Österreich und der Bundesrepublik Deutschland. i. Bodenanalysen. Flora 168: 379-395. sasse F. (1979b): Untersuchungen an Serpentinstandorten in Frankreich, Italien, Österreich und der Bundesrepublik Deutschland. II. Pflanzenanalysen. Flora 168: 578-594. ter braak C.J.F. (1986): Canonical correspondence analysis: a new eigenvector technique for multivariate direct gradient analysis. Ecology 67: 1167-1179. ter braak CLP. (1988): CANOCO. A FORTRAN program for canonical community ordination by [partial] [detrended] [canonical] correspondence analysis, principal component analysis and redundancy analysis (Version 2.1). Upgrade to version 3.1. GLW, Wageningen. van der maarel E. (1996): Vegetation dynamics and dynamic vegetation science. Acta Bot. Need. 45: 421-442. vergnano Gambi O. (1992): The distribution and ecology of the vegetation of ultramafic soils of Italy. In: roberts B.A. & proctor I. (eds.), The ecology of areas with serpentinized rocks, A world view, Kluwer, Dordrecht, pp. 217-247. Received 14 April 1997, revision received 5 January 1998, accepted 19 January 1998