Bryophyte vegetation on cliffs and screes in Western Värmland, Sweden

(1)

(2)

(3)

ACTA PHYTOGEOGRAPHICA SUECICA ₈₆

EDIDIT SVENSKA V AXTGEOGRAFISKA SALLSKAPET

Bryophyte vegetation on cliffs and screes in Western Viirmland,

Sweden

Sven Fransson

UPPSALA 2003

(4)

ISSN 0084-59 14

Editor: Erik S jogren

Technical editor: Marijke van der Maarel-Versluys

Cover illustration: View of GettjamskHitten. Woodcut by Sven Schtitzer-Branzen.

Edidit:

Svenska Vaxtgeografiska Sallskapet Villavagen 14

SE-752 36 Uppsala Lay-out: Opulus Press AB

Printed in Sweden by Fingraf Tryckeri AB , Sodertalje 2004

(5)

Abstract

This study aims at providing increased knowledge of bryophyte flora, communities and habitat conditions on cliffs and screes in south-western Vann.Iand, Central Sweden. Field work was done in 1983- 1 995 . Bedrock, geomorphology and climate of the area were described. As to phytogeography, the area includes the border between the Boreal and the Boreo

nemoral (Hemi-boreal) zones ('limes norrlandicus'). A general survey of geomorphology and hydrology of steep hillsides is given. Cliffs and screes are seen as one functional geo

morphological unity, cliff-and-scree formation (syn. cliff-and

scree).

Vegetation records are of two kinds: species lists from 134 localities and releves of 360 small sample plots. They are sepa

rately treated as to classification, ordination and habitat condi

tions. Twenty habitat factors or conditions are investigated for the localities, 25 for the small sample plots. The mutual correla

tions, 'links', of these habitat factors are given, primarily to detect nonsense correlations. Classification is obtained mainly with the program TABORD, ordination with the program CANOCO. Ellenberg' s (Di.ill' s) Central-European indicator val

ues are used in interpretation of ordination axes. For the species lists of localities, only presence/absence is recorded. For analy

sis of the small sample plots , a geometric series is used as cover degree scale: 1 = < 1/32, etc. to 6 = 112-1 (a modification of the Hult-Sernander-Du Rietz cover scale). Categories of diagnostic species used are characteristic, differential and connecting spe

cies. Western, eastern, southern and northern test groups of species are used to test distributions and connections of phyto

geographical elements.

Three types of cliff-and-screes are revealed, the Plagiopus oederiana, the Cynodontium and the Rhabdoweisia crispata types. Their floristic compositions , habitat demands and geo

graphical occurrences are treated. Four ordination axes are pro

duced. Correlations for all four axes vis-a-vis the environmental factors studied and the indicator values according to DUll are reported. The first two axes are clearly interpretable: the first one is correlated mainly with a poor-rich gradient, the second with a south-to-north exposure gradient. Moisture and temperature are involved in the latter axis , but also here pH was a strong factor.

In the weak axis 4, a southern-northern geographical gradient seemed to be involved. Positions on axis 1 and 2 of localities, cliff-and-scree types and a number of species are reported. The influence of bedrock is discussed. The connections of a number of species to south-to-north exposure by means of average values for all localities of species in question are discussed, as well as relations to a west-east exposure gradient.

The classification of releves revealed 14 clusters, put to

gether into five main clusters: A. Barbilophozia attenuata main cluster: 1 Barbilophozia attenuata cluster; B . Andreaea rupestris main cluster: 2 Andreaea rupestris cluster; 3 Grimmia curvata cluster; C. Barbilophozia barbata main cluster: 4 Hedwigia ciliata cluster, 5 Antitrichia curtipendula cluster, 6 Hypnum cupressiforme cluster; 7 Hypnum andoi cluster; 8 Isothecium

myosuroides cluster; D. Pseudotaxiphyllum elegans main clus

ter: 9 Diplophyllum albicans cluster; 10 Bartramia pomiformis cluster; 1 1 Racomitrium aquaticum cluster; E. Plagiopus oederiana main cluster: 12 Neckera complanata cluster; 1 3 Amphidium mougeotii cluster; 1 4 Neckera crispa cluster.

Floristic compositions and habitat factors are presented in the text, in cluster tables and in two separate survey tables. At ordination, four axes are produced. Correlations for all four axes vis-a-vis the environmental factors and the indicator values according to DUll are reported. Two axes are interpreted: axis 1 is mainly correlated to a poor-rich gradient, axis 2 mainly to shade from cliff structures. Positions on axis 1 and 2 of sample plots, clusters and most species are reported.

Connections with latitude and altitude are discussed for some species by means of average values for all localities of species in question . As to the indicative values of DUll, applied to the species treated, the area is, on average, indifferent between oceanity and continentality; as to the southern-northern gradi

ent, 'KUhlezeiger' according to Dtill is the most frequent group.

Distributions within the area of western, eastern, southern and northern flora elements are discussed in formulae of the number of species of each test group as well as quotients western I (western+ eastern) and southern I (southern+ northern) in each locality. An attempt to connect these figures with climatic fac

tors is made but gave rather weak correlations. Occurrences of species in relation to the limes norrlandicus are discussed. It is pointed out that the bryophyte flora on cliff-and-screes is in this respect principally interesting. The reason is that, in contrast to radicant vegetation, it is almost independent on .soil conditions adjacent to the former highest coastline, which altitudinally nearly coincides with the limes norrlandicus. It seemed that many distributions are still referable to this border.

Clusters were found to have a low homogeneity, partly due to dynamics of this vegetation type. Irrespective of degree of coexistence, they are regarded as communities. The importance of number of releves for characterization of clusters and reliabil

ity of ordinations is discussed. Successions are not studied, which is a deficiency in the study. Generally, cliff structures appear to give better protection for continuity-dependent species than canopy . Some species that are threatened on epiphytic habitats in southern Sweden may have refuges on epilithic substrate farther to the north. Relations between southerly

northerly exposure vis-a-vis latitude and altitude are discussed.

Localities investigated are briefly described, including the most interesting finds.

(6)

1 Introduction

1 . 1 Aim of investigation

side my investigation area. He noted the kind of rock of the substrate on his herbarium labels. Most of his collec- This study aims at increasing the knowledge of bryophyte

flora and communities and their environmental condi

tions on cliffs and screes in an area of south-western Varmland, Central Sweden, as well as to test the possibil

ity of distinguishing different types of such localities on the basis of their bryophyte flora. A corresponding inves

tigation of mire vegetation was made earlier in south

western Varmland, in a somewhat smaller investigation area (Fransson 1972).

The study includes vegetation on real cliffs and screes and thus not on flat rocks or in forests in front of the cliffs or on broad cliff ledges. Epilithic and epigeic communi

ties are studied, of the latter only on leptosols. Neither radicant vegetation (see von Krusenstjerna 1945) on sta

bilized soil has been studied nor epiphytic vegetation.

1 .2 Field work

The field work began in 1983 when species lists of se

lected localities and observations of habitat conditions for the species were made. The field work continued with scattered excursion days until 1994 while the material was gradually worked up. In 1995, I made plant-socio

logical releves and analyses of certain environmental factors of the sample plots. A short report, on Cnestrum alpestre and C. schisti, was published (Fransson 1988a).

1 .3 Previous investigations ofbryophytes on cliffs and screes

The first investigations of the bryophyte flora of cliffs and screes in the province of Varmland were made by Goran Wahlenberg and Claes Gustaf Myrin. The former col

lected, i.a., Tetrodontium ovatum on Brattberget in the parish of Kroppa (Myrin 1832; Wahlenberg 1833). The latter investigated the hills of Ranneberget at bstmark (outside the present investigation area), Gettjamsklatten (no. 118 in this report), and Tossebergsklatten (no. 119) and clarified the differences between vegetation on hyperite (a form of diabase) and granitic orthogneiss. He found, among others, Rhytidium rugosum at Gettjamsklatten.

The geologist H.E. Johansson collected a considerable number of bryophytes on steep hillsides, especially in the areas of hyperites in southern Varmland, but mainly out-

tions are preserved in the Swedish Museum of Natural History in Stockholm. Johansson did not publish this material, but parts of it are used in the bryophyte floras, especially in the later fascicles of the species monographs by Moll er (Moll er 1911-1936). Some information on the bryophyte flora in Varmland in habitats dealt with is also given in Amell & Jensen (1922), H. Persson (1951), Hallingback (1978, 1986, 1987, 1989) and Gustafson &

Hallingback (1994). During recent years the forestry au

thority [Skogsvardsstyrelsen] of Varmland recorded for

est 'key biotopes' by means of 'signal species' (Nitare &

Noren 1992), including a number of bryophytes on cliffs and screes. Partly owing to the fact that these inventories mainly took place after my data treatment, it was not possible to take them into consideration.

Works more directly treating the bryophyte flora and vegetation of cliffs and screes in Sweden are few. Exam

ples are especially Nicklasson (1972) and Sjogren (1991).

In von Krusenstjerna ( 1945), numerous bryophyte com

munities from cliffs in the neighbourhood of Uppsala (where higher cliffs are rare) are described. The same author touches upon this subject in von Krusenstjerna ( 1954) and gives a survey of epilithic bryophyte commu

nities in Southern and Central Sweden in von Krusenstjerna (1965). H. Persson (1936) investigated bryophytes on Omberg, and considerable material has also been pub

lished by Martensson (1955, 1956a, b). Epilithic commu

nities, although not from what can be called cliffs and screes, were treated by Waldheim (1944b) and Sjogren (1964). Jim Lundqvist (1968) treated vascular plants, bryophytes, plant communities and ecological conditions on steep hillsides in Pite Lappmark.

From Norway, we .have St�rmer ( 1938) and J onsgard

& Birks (1993), and from Finland Kotilainen (1929), Pesola (1934), Tuomikoski (1939), and Koponen &

Suorninen (1965). Kotilainen (1960) reviewed botanical research on cliffs in Finland.

(10)

2.1 Situation, delimitation and size

5 10 50 km

--�==�--===---

Fig. 2.1. The province of Varmland, location of the area investigated and isohyps of 200 m a.s.l. After Hard av Segerstad (1952).

(11)

Bryophyte vegetation on cliffs and screes in Western Viirmland, Sweden 9

2.2 Bedrock

The main part of the area belongs to the Southwest Scandinavian Province of the Baltic Shield and its West

ern Gneiss Segment (Th. Lundqvist 1994 p. 16). In the N .E. also the Mylonite Zone and a narrow strip of the Eastern Gneiss Segment are included. The bedrock of the province of Viirmland has been remapped (Lundegardh, et al. 1992, Lundegardh 1995; Gorbatschev 1998; Lindh 1998).

The main rocks westwards from the Mylonite Zone are

Amru.

granitoids, composed of tonalite and granodiorite.

Large sections nearest to the Mylonite Zone consist of granitic and granodioritic gneisses of various origin, often banded and migmatized. Furthest to the SW, the granitoids hold bodies of the Blomskog Granite (Lindh 1977), a Bohus granite type, and other granites, younger than the dominating AmaJ. granitoids.

Westwards from Lake Glafsfjorden, the Gillberga Synform (earlier named the Gillberga Syncline, in Swed

ish 'Gillbergaskalen', Magnusson 1929) is situated, a large structure including the Glaskogen Nappe Complex (Gorbatschev 1998). Peripherally, there are mainly felsic volcanics and granitoids belonging to the Amal group, centrally also granites and bodies of metagabbro and metadiorite. The bedrock to the SW of the Mylonite Zone was difficult to interpret. For detailed presentations of these problems, see Lindh ( 1998) and Gorbatschev ( 1998).

The rocks of the Mylonite Zone (see also Lindh 1974) were massively deformed by the Sveconorwegian orogenesis about 1000 million years ago. Dominating bedrocks are two zones of mylonite, the protholiths of which are unidentifiable. The bedrock between them con

sists of mylonitic deformed orthogneiss, in which some bodies of hyperite (a coarse-grained form of diabase), partly metamorphic, occur.

The Eastern Gneiss Segment consists of granitic orthogneiss. Large bodies of hyperite are common, con

stituting the hyperite hills that are well-known for their rich flora (Myrin 1832).

These are the fundamental features of the bedrock in the area. A quick glance at the map of the bedrock in the National Atlas of Sweden (Th. Lundqvist 1994 p. 32), scale 1: 1 250 000, reveals that bedrock geology to the W of the Mylonite Zone is very complicated; the area has perhaps the most complex bedrock in Sweden. The im

pression is strengthened by the new, more small-scale ( 1:250 000) map. This complicated structure can be found all the way down to the smallest of scales: the kinds of rock alternate in many cases from dm to dm (A. Lindh pers. comm. in Fransson 1988a).

This lack of conformity was considered during the mapping of bedrock in Vfumland by the Geological Sur

vey of Sweden (SGU). Whereas the normal standard for bedrock maps on scales around 1 : 200 000 in Sweden is

one observation per km2, about three, on average, are made in Viirmland (Lindh pers. comm.; Gorbatschev 1998 p. 277). It was not possible to reproduce bodies < 300-400 m2 on the map and consequently bodies of basic rocks and pegmatites are underrepresented (Gorbatschev 1998 p.

277, 313). However, no scale for bedrock maps could reproduce the real conditions and consequently the new bedrock map is also an imperfect tool for looking into connections between vegetation and bedrock.

With regard to botany, the content of calcium is most interesting. For a classification in this respect, see Section 4.3 .1 . , point 10. Connections between bedrock and bryophyte flora are discussed in Section 5.7.

2.3 Geomorphology, altitude

In a classification of relief types of Sweden (Lidmar

Bergstrom 1994 pp. 46, 54), the investigation area be

longs to the undulating hilly land with irregular valleys.

The Sub-Cambrian peneplain is traced fragmentarily ( op.

cit. p. 44), but most of the rock is Precambrian basement.

The undulating hilly land is broken by some faults and lineaments of regional and local importance. Lindh ( 1998) and Gorbatschev (1998) comprehensively treat the exist

ence of faults, deformation zones and lineaments in the area. Fault or thrust zones surrounded by schistosity are marked on the bedrock maps. In nature, faults in recent time are represented by more or less deep and broad valleys and lakes (Gorbatschev op. cit. pp. 339-340).

In the area studied, more or less vertical rocky hill

sides made or influenced by glacial erosion are roches moutonnees, flyggbergs and lakes of rock basin type (Rudberg 1987 p. 112, Fig. 4.4-6). Roches moutonnees are too small to be of interest in this connection. Flyggbergs (see also Rudberg 1954 pp. 318-323, 425-426) are similar to roches moutonnees, but have relative altitudes of 50-

lOO m or more. Like roches moutonnees, they have a steep rock -side in a more or less lee-side position to the direc

tion of the ice-movement. Lakes of rock basin type indi

cate strong influences by glacial erosion; they are U

valleys to a greater or lesser degree.

Cliffs and screes occur most frequently in the south

western parts of the area but time limitations required that even rather good localities there had to be left without investigation. In contrast, it was difficult to find good objects for the investigation in the northernmost parts of the area. Probably several cliffs are hidden under the thick moraine cover there. In the southern and western parts of the area, the moraine cover is considerably thinner or non

existent.

It is difficult to find a measure of the magnitude of precipices; the largest in the area, and possibly in the whole of Viirmland, is that of no. 61, Rankesedsnipan.

The altitudinal differentiation of the area is roughly

(12)

illustrated (Fig. 2.1) by the isohyps of 200 m a.s .l. The formerly highest coastline varies in the area approxi

mately from 175 m a.s.l . in southern parts to 200 m in the north .

2.4 Climate

The meteorological stations in the area are too sparsely dispersed to be a good base for investigations of correla

tions between climate and vegetation within a relatively restricted area like the present one. Even the values of individual stations are to some extent unreliable - e.g. the moving of a station over a relatively short distance may cause considerably changed values . Even more unreliable are values extrapolated between stations . In spite of this , I used such values for, e.g. correlation analyses . However, the relatively low degree of reliability of the figures must be kept in mind. - When drawing 'isolines', I have as far as possible transferred suitable ones from the works cited, otherwise I have drawn them myself.

2.4.1 Temperature

Temperature conditions are described according to Eriksson (1982) and concern the period 1951-1980. The values refer to real levels (i.e. not transformed to sea level) .

Annual temperature mean values (Fig. 2.2) are highest (> 5.0°) in the SW and in the lowlands around the lakes Glafsfjorden and Varmeln . The lowest values ( < 4.0°) are found in the northernmost parts.

Annual temperature range, e.g. the difference between mean temperature for the warmest and coldest months, may serve as a thermic indicator of the position along the gradient oceanity - continentality (cf. Tuhkanen 1980).

The amplitude of this parameter within the area is between 21.0° and 23.4 °, mainly extended in the direction SW-NE.

Two areas, however, deviate from this pattern by lower values than the surroundings, one around the lakes Stora Gla and bvre Gla and the districts to the SE of them, the other one around the lakes Rottnen and Kymmen.

Within this annual amplitude, winter temperature mean values (⁼coldest month, January or February) have the greatest differences , 2.5° (-5.2° - -7 .7°). They are distrib

uted along a rather straight south to north line . The differ

ences in the temperature mean values of July are moderate 1.2 °C. They are distributed in the direction SE-NW; the warmest (> 16°C) being the low-lying valleys surround

ing the lakes Glafsfjorden, Varmeln and Nedre Fryken, the least warm (15 .3-15 .8°C) being the westemmost parts.

The distribution of the annu_al temperature range within the area may be related to the fact that major parts of the area - exceptions are the south-westernmost parts - be

long to the ' Local continentality area K 1 ' , according to

Angstrom (1974 p. 53), but are situated in the southern periphery of it, where the continental character is weak.

Only about 50 km to the SW of the investigation area, we have the northern limit of Angstrom's ' Local maritimity area M l'.

2.4.2 Precipitation

Major parts of the area get precipitation within the range 700-800 mm (Alexandersson et al. 1991). An area with<

700 mm extends from the Fryken lakes westwards to Arvika and Amotfors . Precipitation of > 800 mm prevails within a large area extending from the Norwegian border across the parishes of Jarnskog and bstervallskog south

eastwards over the western parts of the raised area of Glaskogen.

After corrections for various measuring deficiencies (losses by wind, evaporation and adhesion), Alexandersson

& Andersson (1995) presented estimated mean values for the period 1961-1990 ('actual annual precipitation'). Ac

cording to these values, the major parts of the area get higher precipitation (800-900 mm) than generally expected and also a deviating distribution (Fig. 2.3). According to Alexandersson & Andersson (op. cit.), this version of pre

cipitation distribution is called 'actual' . The alternative version, mentioned above, I call 'traditional'.

Fig. 2.2. Annual temperature mean values . Measured values according to Eriksson ( 1982). 0 cliffs and screes investigated (see Fig. 2.5).

(13)

T

800

20 km

'---'---'

900

Fig. 2.3 . Annual mean precipitation, 'actual'. After Alexandersson

& Andersson (1995).

The precipitation distribution over the year is also interesting. A continental feature in this respect is a high summer maximum (see Tuhkanen 1980 p. 41). Hard av Segerstad (1952 pp. 21-25), making use of figures in Wallen (1924), illustrated that the Fryksdalen area in this respect is more continental than SW Vfumland.

2.4.3 Humidity

Many formulae expressing humidity have been constructed (cf. e.g. Tuhkanen 1980 p. 23 ff.). For Swedish 'normal terrain' , the humidity index of 0. Tamm ( 1959) has often been applied. It is defined as the difference between annual mean precipitation and a calculated evapo-transpi

ration. As a rule, precipitation increases with altitude, whereas the temperature-dependent evapo-transpiration decreases. Thus, the humidity index increases still more with altitude than precipitation does. An attempt to give a more detailed picture of the humidity of parts of SW Varmland (the afore-mentioned area for my mire studies) is found in Fransson (1972).

Eriksson ( 1986) calculated the humidity during the growing season as the difference between precipitation amount (corrected values for the period 1951-1980) and calculated values for 'real evapotranspiration' (in contrast to potential evapotranspiration, e.g. the transpiration when vegetation and soil are optimally supplied with water).

The distribution in the area is given in Fig. 2.4.1t shows a

50

Fig. 2.4. Humidity during growing season. After Eriksson (1986).

large-scale, but nevertheless interesting, version. Values within the range 150-200 mm, prevailing in an area in the west, imply very high humidity for Swedish conditions, existing only in the westernmost parts of southern Swe

den and along the Norwegian border from N Varmland to S Lapland. Values between 50 and 0 mm, existing just outside the south-eastern border of the investigation area, may be considered as normal values of Central Sweden.

2.5 Phytogeography

The area includes a geomorphological and phytogeo

graphical border zone, often according to Sernander (see e.g. Fries 1948 p. 56, Fransson 1965) referred to as the limes norrlandicus. Phytogeographically it distinguishes the Boreal zone (its Southern boreal subzone) from the Boreo-nemoral (Hemi-boreal) vegetation zone (Sjors 1963, 1965, 1999, Fransson 1965).

As to the vascular plant flora, we have a detailed investigation of the plant geographical conditions at this border in Vfumland (Hard av Segerstad 1952). Malmgren (1982) made a corresponding investigation of Vast

manland. In the afore-mentioned parts of SW Varmland, I mapped mire plants (Fransson 1972), including bryophytes.

For the province of Narke, Waldheim (1944a) treated Sphagna. 0kland (1990) treated mires from a phyto

geographical viewpoint in 0stfold and Akershus in Nor-

(14)

way , adjacent to the present study area.

In the zone of the limes norrlandicus, phytogeo

graphical frequency diversities , altitudinal transitions and the former highest coastal line , with its transitions between moraine and marine and lacustrine deposits , roughly coincide. Bryophyte flora on cliffs is interesting in the light of these conditions; it is hardly influenced by distributions of soils, and many cliffs are situated near or on (or actually constitute) the border between lowland and highland. Thus , one may ask whether the border is traceable at all in my material.

A W-E division into a more suboceanic and a more continental (more properly subcontinental) part is traceable, especially in the southern part of the investigation area.

T

2.6 Investigated localities , registration sub areas

Foremost in order to register the finds, the localities are numbered and arranged in parishes and five registration subareas . Parishwise labelling and reporting are tradi

tional approaches in Swedish floristics . The positions of the localities , parishes and registration subareas are given in Fig. 2.5. For names and descriptions of individual localities, see Appendix 1.

<-��--'lYS

119(), ·.,\

� '�'�, ^r��

0 20 km

Fig. 2.5 . Investigated localities (numbers) , parishes (abbreviated) and registration subareas. - Names and descriptions of individual localities are found in Appendix 1 . Registration areas and parishes: The south-western parishes (nos. 1-22): Svanskog (Svan), Sillerud (Sill), Blomskog (Blom), Silbodal (Silb), Tocksmark (Tocks), Karlanda (Karla). The south-eastern parishes (nos . 23-47): Gillberga (Gill), Langserud (Lang), Borgvik (Borg), Stavnas (Stav), Varmskog (Varm), Frykerud (Fryk) , Boda, Stora Kil (St Kil) , Brunskog (Brun) . The central parishes (nos. 48-75): Glava, Alga, Arvika (Aa), Josse Ny (J Ny) , Hogerud (Hog, no locality investigated) . The north

western parishes (nos. 76-t'l l): Ostervallskog (Osterv), Jamskog (Jam) , Skillingmark (Skill) , KOla, Eda. The north-eastern parishes (nos. 1 12-134): Vastra .Amtervik (V .Amt) , Mangskog (Mang) , Sunne, Gunnarskog (Gunn) , Grasmark (Gras), Lysvik (Lys), Fryksande (Fryks) .

(15)

3 Morphology and hydrology of cliff-and-scree formations

The morphological formations, the bryophyte of which this study deals with, are± vertical rocky hillside , cliffs , and debri fallen from them, screes . I con ider the combi

nation of cliff and scree to be a functional unit, a cliff-and

scree formation (cf. cliff and scree vegetation , Jim Lundqvist 1968, p. 84), sometime written cliff-and-scree for short (pl . cliff-and-screes). For di cus ion on an appropriate term in Swedish for the concept of cliff-and- cree formation, see Du Rietz (1954) and Wi trand (1962).

I propose as a Swedi h term klippa-talu -formation . In this concept I do not include the more or le horizontal , forested top o f the mountain (cap , calotte , plateau , see Jim Lundquist 1968 p. 17). Admittedly, this top influence the cliff and the cree geo-morphologically by supplying the denudation proces es with material . However, when this happens it has become a part of the cliff-and- cree forma

tion . Ecologically however, it influence the cliff-and

scree by water upply . - Alternative term are possible (see especially Rapp 1960, 1961 and Jim Lundqvi t 1968), but to avoid mistakes, I have chosen not to use them .

I have consistently tried to use cliff as a term of morphological formation , and rock (rocky) a a term of matter (bedrock); cliff structure concern morphological forms of the cliff, and not (bed)rock structure or texture . The cliff consists of several morphological elements , cliff structures. It may be a whole, nearly continuous urface, rock-face (face) . Generally, however, the cliff consists of several rock-faces . The faces may be vertical overhanging (overhang ) or sloping . Different face of a cliff at different levels are separated by ledges, often

Plate 1. No. 6 1 Rankesedsnipan is the mightiest cliff-and- cree for

mation of the area tudied. The steepest part i about 600 m broad and reaches about 1 50 m above Lake Ranken. The light part dis

cernible in middle of the picture are traces of a rockslide which oc

curred in 1 992- 1 993 . It stretches from the upper part of the scree down to below the lake surface.

30.9 . 1 998.

traver able, making investigations of the cliff flora easier.

The e ledges are named terraces , if they are broad enough for forest plant communities , belonging to a± horizontal and podsolized ground , to grow there, shelve if they are broader than 5 cm, borders if till narrower. As to the angle with the rock-face , the ledges are horizontal or loping inwards or outwards . Shelves and borders are e pecially common on chistou and gneissic rocks. Soi l i accumulated o n shelves and borders too, but not a s deep and podsolized layer (they are leptosols) .

Large , nearly completely mooth rock surface exist especially where the surface is parallel with the structure layers of± gneissic rocks . Homogeneous rocks often have a round-polished surface .

In the cliff, cavities may occur. They are called cave if they have a fairly narrow entrance , in other case crevices. Cavities may have roofs , walls and floor . The largest cave in the area is found in no. 113 Gylterudsbergen (Appendix 1).

Fissures in various directions are common . They ac

cumulate soi l and give footholds for bryophytes. If the fissure are± horizontal or vertical , this may imply differ

ence in conditions for mos es (e .g . for water-flow ) . If the urface of the fractures have become removed from each other, we have cases of chinks if they are only some cm broad , or clefts if they are broader. Very broad fissures are called gates (Swedish ' kurugator' , in accordance with 'Skurugata' in northern SmiHand) . Gates that are a few metres wide , directed transversally to the cliff l ine , are found at no . 52 Ramlekleva and 87 Mellanfjallen. -

Acta Phytogeogr. Suec. 86

(16)

Plate 2. Big boulder in the cree of no. 86 Fjall berget and Skutan. Anastrophyllum saxicola and Tetralophozia setiformis often occur in uch habitat . 30.9.1998.

Upper ± horizontal ection of the cliff quite frequently project as a terrace roof over ubjacent layer . They may have been formed a hort thru t or by cracking and

ub equent removal of interjacent strata.

The mechanical weathering causing denudation of cliff-and-scree i mainly by freeze-and-thaw proces e in combination with loping proces e . Among the latter, rockfall (more or les free fa11 s of rock particle ) and rock I ides (rapid, liding ma -movement ) are important (Rapp 1961 pp. 7 4 97 1 1 8) . The peed of the e proce es depend to ome extent on the kind of bedrock . Rudberg (1954 pp. 239 , 422) gives an order of resi tance of the bedrocks of CaJedonidian rock , but find no ob ervable differences between Archaean rock . However, among the rocks of the Caledonides, migmatite gneiss is more easily weathered than granite according to Rudberg , and Jim Lundqvi t (1968 pp. 1 5-1 6) maintain that the me

chanical weathering i more rapid if the bedrock is hetero

geneous or if the cliff consist of different kinds of rock . Especially on smal l dry cl iffs , consi ting of hard , homo

geneous rocks , the fall of debris is slow . In these cases , it eem that not so much ha happened since glacial ero

sion . In larger cliffs , strong breaking-down processes are working, especially during springtime. No . 21 'Varbrott

fjallet' which mean ' pring-break mountain' ha urely Acta Phytogeogr. Suec. 86

got its name becau e of observations of such phenomena and at no. 61 Ranke ed nipan , there were recent rock

tide ; one relatively large and two maller ones. A mall rockslide recently happened in no. 1 1 0, the eastern side of Norealven (See Appendix 1 ) .

Water can apparently more easily percolate and aggra

vate the freezing process between the rock layers in het

erogeneous rocks than in homogeneou one . However, a prerequi ite is, one may presume , that the layer form any angle with the urface of the rock-face; most favourable is presumably an inward lope . Rock-faces parallel to the tructural strata of the bedrock have, at lea t a a rule, an acidicolou bryophyte flora quite l ikely in many cases depending on slow weathering, but other reasons cer

tainly contribute . - The mechanical weathering, and the peed of it, affect the vegetation by offering new open urface for colonization free from competition, by creat

ing more variable habitat condition , and by increasing the chemical weathering by means of uncovered fresh

urface and an enlarged total attack area.

Water-flows on the cliffs are exterior or interior. Exte

rior flow emanate from precipitation directly on the cliff urface or from surplu urface water on the top . Interior flows emerge through fi sures. A they have pas ed through deeper, less leached layers they often provide prerequi

site for a rich flora (see e .g . Ander son & Birger 191 2 , pp . 64-65 ; D u Rietz 1954; Jim Lundquist 1968 pp . 20, 61 ff.). Slope of heterogeneous rock are genera11y better watered than slopes of homogeneou rocks , as they create more paths for seepage water (Jim Lundqvi t 1968 p. 1 7 ).

A good watering gives better fro t activity , it contribute to a more rapid weathering in heterogeneous rocks . - All water-flow have potential for nutritional enrichment by means of evaporation .

Plate 3 . A small cliff-and- cree formation. Part of no. 96.

Rammberget at Lake Yadljungen. 30.9.1998.

(17)

Bryophyte vegetation on cliffs and screes in Western Viirmland, Sweden 1 5

Plate 4. Traces of a mall, recent rock- lide, 6 m broad and 8 m long (vertically 4 m) and in addition ome boulders further down. - Rock- tides offer new ubstrate for bryophyte to colo

nise. No. 110 Norealven 4.11.1998.

In periods of rain or now-melting , water percolates, rather rapidly, into fissure of the rock, oozing out from roof and wall of cavities and overhangs . Water pa es also over to interior rock- urface from edge of over

hangs and flow over them . Also overhanging urface retain parts of this water, as adhesive powers have a

tronger influence on the thin water-film than gravity . As water from higher level exert pressure , the water dis

per es into nooks and crannie far away from rain or constant flow . Under a roof, the nutritional effects of this water-flow are not counteracted by the leaching effect of precipitation , the con equence being that cliff- ections under a roof in many ea es have a more basicolous bryophyte flora than rain-expo ed cliff-part .

In very wet conditions, exterior water runs or drops from the edge of overhang to the ground in front of the cliff. On the point or zone of fall , the drop zone , the bryophyte flora i influenced .

It i not ea y to decide the extent to which there exist bryophyte stands whose available water exclu ively ema

nates from the humidity of the atmo phere . I have been surpri ed to find how far into crevice and caves protected from direct rainfal l the water reache after period of rain and melting away of snow . Such water-flows on rock- urface far from the reach of direct rain, even of hort duration or on rare occasions, must have a great influence on vegetation - besides upply of water - by additions to the air-transported nutrients. Such water-flows are best seen when fro t has created icicle on the roofs or frozen waterfalls on the walls of cavities .

By means of sloping proce ses , the mechanical weath

ering on the retreating cliff has generated scree . If the mechanical weathering and retiring of the cliff are uni

form alongside the cliff a imple cree slope will be

Plate 5. During water-rich period , water come oozing al o in ites under roof. Here icicle formed by the fir t autumn fro t how up the water-flow . No. 110 Norealven 4.11.1998.

formed (cf. Rapp 1960 p. 4). If the e proces e are fa ter in some sections, rock fall funnel will be formed in the erosion part, and coned-formed cree slopes in the accu

mulation part (Rapp op . cit . p. 5). Small but typical examples are to be seen on no. 98 Kronefjallet and no. 1 1 8 GettjarnskHitten . Narrow rock fall funnels often seem to have an intere ting bryophyte flora; at both the two local i

ties mentioned above Plagiobryum zierii i found ju t at uch rock fall funnel . - For the purpose of elf-preserva

tion visitors should bear in mind the ri k that they may tart a rock fall at the e funnel , especially if they are wet . In the cree slope , the rock debri i fractionated with di tal ly increa ing particle ize . At the cree top, just in front of the rock-face , there is a zone with fine cree (Sw . 'finur' ). E pecially at a simple cree lope there is often another fairly practicable path for walking along ide the cliff. At the scree base there is a border of the largest boulders , in many case very large . At lea t two specie , Anastrohyllum saxicola and Tetralophozia setiformis, pre

fer scree boulders to the firm cliff.

The whole of the material below a cliff i not alway debris from the cliff it elf but may be the re ult of accu

mulation after transport from afar mo t commonly gla

cial tran port . This might explain why in ome cases, there is a rich vascular plant flora below a cliff with poor bryophyte flora (e .g . no. 39 Storhojden) or vice versa, i .e . a poor v a cular plant flora below a cliff with rich bryophyte flora.

Smaller cliff-and-screes have a less complete struc

tural hape than de cribed .

Acta Phytogeogr. Suec. 86

(18)

4 . 1 Introduction

According to the aim of the investigation, two types of sample areas were studied: whole cliff-and-screes and small sample areas within them. Both were computer

processed separately, and the species lists of the cliff-and

screes were then treated as releves (as, in a proper sense, they are). When risks for mistakes exist, I call the material of the cliff-and-screes 'cliff material' , 'cliff analysis', etc., and the material of the sample plots for 'releve material' , etc. The ordination axes are called 'DCA 1 cliffs' , 'DCA1rel', etc. With regard to environmental factors, some are suitable for the cliff analyses, some for the releve analyses, and some for both. Consequently, there are two different lists of environmental factors, which might initially be confusing, but result in a better environ

ment analysis, as the DCA-analysis of the cliffs gives alternatives and complements to the treatment of the releves. Species stocks of whole cliff-and-screes may partly be controlled by other factors than small sample plots (cf. for mires, 0kland 1990 pp. 108-109) . The ordi

nation of cliffs might also give supplementary informa

tion on relatively rare species with a frequency in the sample plot material that is too low for reasonably reliable conclusions.

4 .2 Vegetation analysis

4.2.1 Selection and inventory procedure of the cliff

and-screes

Most large cliff-and-scree formations in the area have been investigated, attempts being made to get the distribu

tion over the area as equal as possible. The majority of the investigated localities were visited at least twice, ·others only once. The decision to make additional visits mainly depended on an assessment of the chances to find more species. On the whole, it is difficult, or even impossible, not to have omitted some species even after very thorough inventories. The three-dimensional structures of cliff

and-screes often contain systems of nooks and crannies that are nearly impossible to survey. In many cases, cer

tain parts are difficult (and even dangerous) or impossible to reach without mountaineering equipment, and such outfits were not used. Naturally, this resulted in different degrees of incompleteness of the species lists. Some lo

calities with clearly incomplete records have been ex

cluded.

The inventory procedure was as simple as possible: in the localities bryophyte species were recorded without quantitative differentiation.

Very common species were not consistently recorded (mainly non-exclusive cliff-and-scree species). However, I have treated such incompletely recorded species if the records of them could be considered as representative.

4.2.2 Analysis of the sample plots

I could not find any practicable way of sampling small sample areas at random in a strict sense, and neither could I find a system comparable with randomness (see e.g.

Greig-Smith 1957). Both cliffs and screes (the boulders) are three-dimensional. When applying any kind of ran

dom method, one often discovers that the sample plot lands within the rock, on places impossible to reach by others than mountaineers, on bryophyte-bare plots, half

way between two rock-faces, etc. Probably thousands of releves would be needed to get less common communities represented. Systematic plotting within homogeneous stands using some kind of regular pattern is not a practica

ble method either, as uniform stands are so small in this diversified vegetation; usable stands would give an in

complete picture of the vegetation.

Consequently, I used another approach. First, I ac

quired a general view of the vegetation in a locality. After this 'reconnaissance' (Cain & Castro 1959 pp. 105-106, Westhoff & van der Maarel 1973 p. 631), I selected uniform stands that together represented as well as possi

ble the whole variation of the vegetation within each locality, and finally the whole area. To reach uniformity, I did not require a special shape or size. Concerning size, I accepted a variation between 1 and 0.1 m2, in excep

tional cases still smaller, on account of the cliff-and-scree vegetation often covering very small areas of different shapes, e.g. alongside fissures or in small crevices. To reach a (qualitative) representation of the vegetation, I endeavoured to include representation of different geomorphological structures and, above all, to include as many species as possible into the final material .

Of a total of 134 investigated localities, I made releves in 27 of them. I tried to get them uniformly distributed both geographically over the area and with regard to types of cliff-and-screes. However, unfortunately, none of the ten cliff-and-screes of the Cynodontium type (Section 5.1) is represented.

A consequence of this arbitrary sampling procedure - in general accepted in phytosociology (see e.g. Westhoff

(19)

& van der Maarel 1973 and Malmer 197 4) - is, of course, that the sociological material is not quantitatively repre

sentative. Rare species and communities are over-repre

sented, and parts without bryophytes, which are common, are left out. The subjective sampling must be kept in mind when discussing the numerical analyses, not at least the correlation analyses. It is also important to take into consideration the mutual 'links' between environmental factors investigated (Section 4.4).

In the field the degree of cover was estimated in percentages, whole numbers and < 1 %. In that way, it was possible to later choose among scales for degree of cover (or to compare effects of different scales of cover values on classification, a prospect not realized, however).

When choosing among coverage scales, I tried to find a position between - on one hand - only presence-ab

sence, when cover degree has no influence on the result, and - on the other hand - true percentage, when cover degree will be of primary importance. I wanted especially to diminish the influence of differences in cover in posi

tions > 50 % . A logarithmic scale of some kind (see Jensen 1978) appeared to be acceptable, but with a risk that cover would become a too insignificant factor. Fi

nally, I decided to choose a geometric series: 1 = < 1/32, 2

= 1/32-1/16, 3 ⁼1116-118, 4 = 118-1/4, 5 = 114-112, 6 = 11 2-1. When transforming from percentage, I applied: 1-3

% = 1 , 4-6 % = 2, 7-12 % = 3, 13-25 % ⁼4, 26-50 % ⁼5, 51-100 % = 6. This scale may be seen as a variant of the scale of Hult-Semander-Du Rietz, modified by introduc

ing a degree below 1/16 and, as a consequence, by shifting the cover degrees. (In principle, a transformation from percentage to the scale of Hult-Sernander-Du Rietz is not fully possible: which degree of cover have 25 and 50 % ?)

4.2.3 Categories of diagnostic species used

The categories of diagnostic species used are character species, differential species and connecting species. Con

cerning the former two, I refer to Westhoff & van der Maarel (1973). Connecting species is a synonym (pro

posed by E. van der Maarel pers. comm.) for binding species (see Fransson 1972 and literature cited therein).

With regard to the term differential species, there are logical reasons for it to be complemented with an equiva

lent term indicating similarities, thereby enabling both differences and similarities to be compared, not only differences. True, connecting species are often character

species of communities of higher rank, but not always. At non-hierarchical , multidimensional systematization, this term should be of great value.

4 .3 Environmental factors studied

4.3.1 Environmental factors of the cliff material For numerical analysis of the environment of the whole cliffs, I compiled scores for the following factors (or in certain cases conditions) for each locality:

1 . Latitude, as distance (km) to the N from the southemmost locality.

2. Longitude, as distance (km) to the E from the westemmost locality.

3 . Altitude (m) at the foot.

4 Altitude (m) at the top.

5. Height (m) = top height minus foot height.

6. Length (m) .

7 . 'Steep value' = number of equidistances (interval 5m) omit

ted.

The topographical map of Sweden, at scale 1 :50 000, (omitted equidistances at steep hillsides for want of space due to crowding.

The number of equidistances omitted gives to a certain degree an objective measure (at least not depending on the author's subjec

tive assessment) , of the proportions of the very steep parts of cliff

and-screes, possibly better than relative altitude.

8 . Exposure, ^ofrom the N (gons) . 9. Exposure, ⁰from the W (gons).

10. Bedrock classes according to content of calcium.

Bedrock or predominant bedrock of the localities, according to the petrographical map, scored as:

5: Sedimentary rocks belonging to the Dalsland group; carbonate- rich, sedimentary rock;

4: Hyperite; schists belonging to the Amal group (Glava schist);

3: Tonalite and granodiorite; metagabbro and metadiorite;

2: Granitic and granodioritic gneisses; granodiorite and granite;

1 : Granitic gneiss; gneissic granite, red-greyish red; orthogneiss, granitic; predominating felsic volcanics; porphyritic granite;

Blomskog granite; augengranite of the Strom type, Strom gneiss;

usually gneissic granite, normally red-greyish red; very coarse

grained, granitic augengneiss. - Strongly migmatized, inhomoge

neous gneisses and mylonite, represented on four localities, were excluded. -In this classification, I referred to Eklund ( 1 953).

1 1 . Annual temperature mean value (Section 2.4.1) with the scored intervals: < 4 ° = 1 score; 4-4.5° = 2 scores; 4.5-5° = 3 scores; > 5° = 4 scores .

12. Annual temperature range (Section 2.4.1), with the scored intervals: < 21 ^o= 1 score; 21-22° = 2 scores; 22-23° = 3 scores; > 23° = 4 scores.

1 3 . Temperature mean value for coldest month (Section 2.4^.1) with the scored intervals: < -7° = 1 score; -7 to -6°= 2 scores; > -6° = 3 scores.

14. Temperature mean value of July (Section 2.4.1) with the scored intervals: < 1 5 .5° = 1 score; 1 5 .5 - 16° = 2 scores; 1 6 - 16.5° = 3 scores; > 16.5° = 4 scores.

1 5 . Annual mean precipitation, 'traditional' (Section 2.4.2), with the scored intervals: < 700 mm= 1 score, 700 - 800 mm

= 2 scores, > 800 mm = 3 scores.

(20)

16. Annual mean precipitation, 'actual' (Section 2 .4.2), with the scored intervals: < 800 mm = 1 score; 800-900 mm = 2 scores; > 900 mm = 3 scores .

17. Humidity during growing season (Section 2 .4.3) with the scored intervals : 50- 100 mm ⁼1 score; 100-1 50 mm = 2 scores; 150-200 mm = 3 scores .

1 8 . Position within or outside the occurrence area of Tricho

phorum caespitosum according to Hard av Segerstad (1952) and Fransson ( 1 972). Within ⁼1 ; outside = 0. Included as a possible measure of humidity .

19. 'Index of rich cliff species' , i.e. on each locality recorded number of 24 selected species , in literature known as basicolous species. Aimed at a measure of the position of the locality on the gradient poor-rich. Strictly , this parameter is not an environmental factor but an indication of the vegeta

tion itself.

20. Cultivation zone according to the zone map ofRiksforbundet Svensk Tdidgiird (Aim et al. 1 991). Aimed at a measure of mildness-harshness of climate tested in practice. In the area, zones 3, 4 and 5 occur, the first-mentioned the mildest. The zone numbers were used at the data treatment.

It is important to keep in mind that only the scores of points 1 - 10 and 19 are based on measurements at each locality, the others are estimated or extrapolated scores.

4.3.2 Environmental factors of the releve material The following environmental factors (or conditions) were measured or estimated:

A. Continuity of forest close enough to influence the sample plot (line 1 in Table 4.2):

4 = PrimevaVprimeval-like forest;

3 = Old forest (> 30 yr);

2 = Young forest (> 10 yr);

1 = Clear-felled area.

B. Directions of exposure (lines 2-5):

Direction of exposure, ^ofrom the N of the sample plot, trans

formed into 10 classes (line 2) .

Main direction of exposure of the cliff-and-scree, ^ofrom the N (line 3), reported in the same manner.

Deviation to the S of the plot from the main exposure of the cliff

and-scree (line 4), calculated as the exposure class of the plot according to line 2 reduced by the exposure class of the cliff according to line 3 . For want of space (negative signs) not reported for the individual releves but as mean value for each cluster (Table 6.16) .

Direction o f exposure, ⁰from the W, transformed into 1 0 classes (line 5 ) .

C. Light (lines 6-7)

Light, shaded by the cliff structure (line 6):

1 = Strong shading (caves, deep crevices, strong overhangs);

2 = Less strong shading (more open crevices or overhangs);

3 = Slight shading (mainly the rock-substrate itself);

4 = No shading (top sides of boulders).

Light, shaded by the canopy (line 7):

1 = Very dense canopy;

2 = More transparent canopy or dense with small glades;

3 = Rather transparent canopy or dense with large glades;

4 = No canopy.

D. Structures

Rock faces: Degree of overhanging (line 8) in ° from the vertical plane, in the following classes:

0 = sloping-0°;

1 = 1 - 1 0° (= almost vertical);

2 = 1 1 -20°, .^{. .}and so on to 9 = 8 1 -90° (= horizontal roof).

Rock-faces: Degree of sloping (line 9), in the same class divi

sion as above:

0 = overhanging -0°;

1 = 1 - 10° ( = almost vertical) and so on to 9 = 8 1 -90° (= horizontal 'floor').

Rock-faces: vertical (± 10°) rock-face (line 10: in classes 110, i.e. presence/absence . Surfaces of boulders are classed in the same manner as the rock-faces .

Shelf (line 1 1), classes 110.

Fissure (line 12) , classes 110.

Boulder (line 13), classes 1/0.

Under roof (line 14), i.e. positions under overhangs and 'terrace roofs' or on cavity roofs.

E. Wetness and water

Wetness quantity estimated (line 1 5), classes 1 -5:

1 = Much less than quantities o f rain only;

2 = Less than quantities of rain only;

3 ⁼Quantities about these of rain only;

4 = More;

5 = Much more than quantities of rain only (permanent flows).

Kinds of water supply:

The plot is reached by rain directly (line 1 6), classes 1/0. - Not exact the same as 'under roof' (see above), as rain may reach positions under roof to windward. This was not satisfacto

rily assessed until releve 68 .

(21)

The plot is reached by interior flows (line 17) classes 1/0.

The plot is reached by exterior flows (line 18), classes 110.

(As the analytical work took place throughout a summer, during both wet and dry periods , it was difficult to correctly estimate both wetness quantity and kinds of water supply. They may be regarded as comparatively unreliable, especially wetness quan

tity . Water-flows also include occasional flows, traceable as shallow grooves on the rock-face.)

F. Numbers of basicolous bryophytes close to the sample plot (line 19).

O = None;

1 = 1 -3;

2 = > 3 species within a radius of 2 m.

G. The proportion of deciduous trees and bushes in the canopy within a radius of roughly 5 m (line 20).

1 = Conifers only;

2 = Conifers dominating;

3 = Conifers and deciduous bushes and trees equal;

4 ⁼Deciduous bushes and trees dominating;

5 = Deciduous bushes and trees only.

H. Indications of mechanical weathering in progress (in rock structure or structure of the rock surface, occurrences of fresh debris , loose particles, movable soil (line 21), classes 0/1 . -Appeared to be a factor sometimes difficult to assess.

I. Substrate

Rock (line 22), classes 110 (presence/absence).

Soil, (line 23), classes 110 (presence/absence).

(The notes on substrate in the tables are not fully applicable to each individual species. Thus , some releves are noted for both soil and rock. Even if only one of the substrates is noted, small liverworts may, for example, grow within stands of other

bryophytes, or, if rock, a small accumulation of soil may remain concealed under the moss cover unless the vegetation is re

moved, which I did not do. Besides, there are frequent border

line cases between rock and thin soil as substrate.)

J. Lichen-covered proportion of the plot, o/o, (line 24) , trans

formed to the classes 0, 1 = 1-10 % etc. to 9 = 81- 100% (thus not according to the cover scale of vegetation analysis) .

K. Proportion of vegetation-less substrate of the plot (line 25 : 'Bare substrate' ) , classes as line 24.

4.4 The mutual correlations of the en vi

ronmental factors

The environmental factors studied have mutual correla

tions, of which several are causal and even obvious (Ta

bles 4.1-2). Some of them are physically coupled, posi

tively as, e.g. altitude at the top and at the foot in the cliff analysis, or negatively as between soil and rock substrate in the releve analysis. These correlations are to be seen as prerequisites for the interpretations of correlations be

tween vegetation and environmental factors, foremost to detect nonsense-correlations. Hence force these correla

tions are called links.

Especially concerning the releves, a further reason to attach great importance to these links is that the sample plots are not sampled at random. An example of the consequences of this fact is that exterior flows are posi

tively correlated to positions under roofs (r = 0.37). The reason is hardly that exterior flows should be more com

mon under roofs than in open positions. In positions under roofs, however, bryophyte vegetation on the whole is more common at flows. Thus, in positions under a roof, flows (both interior and exterior ones) became over-rep

resented and the factors positively correlated. In the cliff

Table 4. 1 . Mutual correlations ('links') of environmental factors, the cliff material. Factors according to Section 4.3.1 .

1 Latitude 2

2 Longitude 0.21

3 Altitude at the foot 0.28 -0.09 4

4 Altitude at the top 0.39 0 0.80

5 Height 0.21 0 . 1 1 -0.25 0.35 6

6 Length -0.13 0.02 -0.26 -0.06 0.29 7

7 'Steep value' 0.15 -0.05 -0.09 0.33 0.66 0.29 8

8 Exposure ⁰from N 0.24 0 . 1 2 0.02 0.18 0.24 0.03 0.36 9 9 Exposure 0 from W 0.1 1 0.04 O.Q7 0.10 0.05 -0.06 -0.03 -0.02 10 10 Bedrock -0.12 -0.07 -0.17 -0.18 -0.03 0.22 -0.03 -0.05 -0.07 l l 1 1 Annual temperature -0.77 -0.10 -0.43 -0.50 -0.16 0.20 -0 . 1 5 -0.36 -0.17 0.12 12 12 Temperature range 0.62 0.41 -0.19 -0.09 0 . 1 3 -0.04 -0.03 0.21 -0.01 -0.03 -0.36 13 13 Temperature winter -0.79 -0.17 -0.22 -0.29 -0. 1 3 0.09 -0.05 -0.32 -0 . 1 8 0.08 0.72 -0.65 14

14 Temperature July -0.21 0.60 -0.50 -0 .47 -0.0 1 0.12 -0. 14 0.02 -0.09 0 0.29 0.24 0.16 IS

1 5 Precipitation 'trad.' -0.16 -0.63 0.43 0.32 -0 . 1 3 -0.08 0.06 -0. 16 0.06 -0.04 -0.01 -0.64 0.19 -0.5 16 1 6 Precipitation 'actual' -0.22 -0.32 0.37 0.37 0.08 0.02 0 . 1 5 O.Q7 -0.04 -0.09 -0.09 -0.42 0.1 1 -0.38 0.41 17 1 7 Humidity 0.35 -0.60 0.34 0.39 0 . 1 3 -0.16 0.26 0.1 1 0.03 -0.14 -0.43 -0.24 -0.22 -0.62 0.58 0.28 1 8 1 8 Trichophorum caesp. 0.04 -0.51 0.46 0.33 -0 . 1 2 -0. 1 1 0.0 1 0.02 0.09 -0.01 -0.24 -0.22 -0.12 -0.66 0.45 0.50 0.41 19 19 Rich cliff spp. 0.16 -0.20 -0 . 1 1 -0.05 0 . 1 1 0.35 0.30 0.10 -0.16 0.32 -0.14 0.04 -0.13 -0 . 1 3 0.08 O.Q3 0 . 1 8 0.22 20 Cultivation zone 0.44 -0.54 0.52 0.5 1 0.05 -0.14 0.14 0 . 1 1 0 . 1 1 -0.10 -0.58 -0.05 -0.43 -0.76 0.50 0.37 0.76 0.56 0.22