Composition and dynamical status of heathland communities in Iceland in relation to recovery measures

(1)

ACTA PHYTOGEOGRAPHICA SUECICA 75 EDIDIT

SVENSKA V AxTGEOGRAFISKA SALLSKAPET

Elfn Gunnlaugsd6ttir

Composition and dynamical status of heathland communities in Iceland

in relation to recovery measures

UPPSALA 1 985

(2)

(3)

ACTA PHYTOGEOGRAPHICA SUECICA 75 EDIDIT

SVENSKA V AXTGEOGRAFISKA SALLSKAPET

Elfn Gunnlaugsd6ttir

Composition and dynamical status of heathland communities in Iceland

in relation to recovery measures

Almqvist & Wiksell International, Stockholm UPPSALA 1985

(4)

Doctoral thesis at Uppsala University 1 985

Suggested citation: Gunnlaugsd6ttir, E. 1 98 5 . Composition and dynamical status of heath/and communities in Iceland in relation to recovery measures. Acta phytogeogr. Suec. 75. Uppsala. 84 pp.

ISBN 91 -72 10-075-3 (paperback) ISBN 91 -72 10-475-9 (cloth) ISSN 0084-59 14

Editor: Erik Sjogren

Technical editor: Gunnel Sjors

Phototypesetting:

Textgruppen i Uppsala AB Printed in Sweden 1 985 by

Borgstroms Tryckeri AB, Motala 1 985

Acta Phytogeogr. Suec. 75

(5)

Abstract

Gunnlaugsd6ttir, E. 1 985. Composition and dynamical status of heathland communities in Iceland in relation to recovery mea

sures. -Acta phytogeogr. Suec., 75, Uppsala. 84 pp. ISBN 9 1 -72 1 0-075-3 (91 -72 10-475-9)

Changes in species composition and cover were studied in perma

nent plots in uneroded areas and deflated ones being restored by protection from grazing , fertilization with mineral fertilizer (NP , NPK, NPKS), and sowing of alien grass seed. The native grass species Agrostis vinea/is, A. capi/laris, Festuca pruinosa and F.

vivipara increased after fertilization as did several small herbs, for example, Cerastium alpinum, Lychnis alpina and Armeria maritima. The first and biggest effect of fertilization only lasts 2-3 years but some minor effects are discernable up to at least 20 years.

The year-to-year analyses of the permanent plots did not reveal any successional trend. On the other hand, fluctuations do occur;

they may be due to fertilization and fluctuating unfavourable weather conditions. A real succession from the vegetation-poor status of deflated areas to fully revegetated ones needs probably 50- 100 years, or more.

In addition, the vegetation in two of the areas was described and classified, and the vegetation of the permanent plots was as

signed to the units obtained. Vegetation types were interpreted as syntaxonomical units according to the Braun-Blanquet ap

proach, and the associations and communities described in this study have been provisionally assigned to higher units in this sys

tem: The open sward assocation Armerio - Silenetum acaulis Ha

dac 1 972 em. is provisionally assigned to the alliance Sedo - Thy-

A grip

Gunnlaugsdottir, E. 1985. Hei�agr6�ur a islandi, tegundasam

setning hans og grMubreytingar me� tilliti til uppgrre3slu. Acta phytogeogr. Suec., 75, Uppsala. 84 pp. ISBN 91-72 1 0-075-3 (9 1 -7210-4 75-9)

Rannsoknir a tegundasamsetningu og pekju groours hafa fari�

fram i fOstum reitum a orkfoka landi sem og ouppblasnu. Fylgst hefur veri� me� gro�urbreytingum, par sem uppgrre�sla fer fram fyrir tilstilli fri�unar, abur�ardreifingar (NP, NPK, NPKS) og saningu innflutt grasfrres.

Spretta innlendra grastegunda svo sem tytulingresis (Agrostis vinea/is), halingresis (A. capillaris), tunvinguls (Festuca pruino

sa) og geitvinguls (F. vivipara) jokst mjog eftir abur�ardreifmgu

na sem og voxtur og fjoldi ymissa smajurta t.d. musareyra ( Ce

rastium alpinum), ljosbera (Lychnis alpina) og geldingahnapps (Armeria maritima). Einnig hafa lagplontur teki� breytingum.

Fyrstu og mestu abur�arahrifm vara a�eins 2-3 ar, en dauf abur�arahrif eru sjaanleg i 20 ar a. m. k.

Arlegar gro�urgreiningar (e. vegetation analyses) i fostu reitunum syna ekki fram a sokn (e. succession) groours ar fra ari. Aftur a moti hafa gro�ursveiflur (e. fluctuations) att ser sta� vegna abur�ardreifmganna og/e�a ohagstre3s arfer�is.

Sokn gro�urs fra gro�urvana, orkfloka landi til algroins �a pvi sem nrest algroins lands tekur 50- 1 ()() ar ooa meir eftir a�stre�um.

A� auki hefur groori tveggja rannsoknarsvre3anna veri�

lyst og hann flokka�ur i einingar gro�urfelagsfrre3i. Einnig er gro�ri fOstu reitanna skipa� i pessar einingar. Grooureiningar

nar (e. vegetations types) eru skilgreindar sem einingar i flokku

narkerfi bygg�u a kenningum Braun-Blanquet og grooursveitum (e. associations) og gro�urfelogum (e. communities), sem her er lyst, hefur til bra�abirg�a veri� skipa� i �ri einingar pessa

mion drucei (B0cher 1 954) de Molenaar 1 976 described from Greenland; the wind-exposed heathland association Racomitrio languinosi - Thalictretum alpini ass. nov. and a related commu

nity are assigned to the Racomitrio - Thalictrion alpini all. nov.

The two above-mentioned alliances are assigned to the order Sedo - Poietalia glaucae de Molenaar 1 976. The grassy heathland associations, namely the Agrostio capillaris - Hylocomietum splendentis ass. nov., the Racomitrio canescentis - Gentianetum nivalis ass. nov. and one related community, are provisionally assigned to the alliance Equiseto - Galion borealis Tx. 1969. The Dryas octopetala - Betula nana community and the Kobresia myosuroides - Sa fix lanata community are provisionally assigned to the class Carici rupestris - Kobresietea bellardii Ohba 1 974.

The species composition of the Icelandic heathland communi

ties deviates from that of related ones described for neighbouring countries. Some Icelandic communities are characterized by combination of species, which are not found elsewhere. This might be explained by different climatic conditions, different bedrock and edaphic conditions, erosion, or any combination of these factors.

Key-words: Phytosociological classification, multivariate methods, permanent plots, vegetation dynamics, deflated areas, fertilization, vegetation restoration.

Elfn Gunnlaugsd6ttir, Institute of Ecological Botany, Uppsala University, Box 559, S-751 22 Uppsala, Sweden.

The Akureyri Museum of Natural History, Box 580, Is-600 Akureyri, Iceland.

kerfis : Gr6�ursveit mela og orfoka lands, Armerio - Silenetum acaulis Hadac 1972 em., hefur veri� skipa� i groourfylki� (e.

alliance) Sedo - Thymion drucei (B0cher 1954) de Molenaar 1 976, sem lyst hefur veri� fra Grrenlandi; gramosapembusveitin

ni Racomitrio lanuginosi - Thalictretum alpini (ny sveit) og skyldu gro�urfelagi hefur veri� skipa� i groourfylki� Racorni

trio - Thalictrion alpini (nytt fylki). IJessi tvo groourfylki heyra til. gro�urlendisins (e. order) Sedo - Poietalia glaucae de

Molenaar 1 976.

Tvrer gro�ursveitir heyra grashei�inni til, p. e. Agrostio capillaris - Hylocomietum splendentis (ny sveit) og Racornitrio canescentis - Gentianetum nivalis (ny sveit) asamt skyldu gro�ur

felagi og hefur ollum veri� skipa� i groourfylki� Equiseto - Galion borealis Tx. 1 969.

Dryas octopetala -Betula nana gro�urfelaginu og Kobresia myosuroides - Salix lanata gro�urfelaginu hefur til brMabirg�a veri� skipa� i gro�urflokkinn (e. class) Carici rupestris - Kob

resietea bellardii Ohba 1974.

islenski heit�agroourinn er mjog frabrug�inn skyldum gro�ri i grannlOndunum. Grooureiningar hans eru einkenndar af hopum tegunda, sem hvergi annars sta�ar vaxa saman. Orsakanna er a� leita i frabrug�inna voorattu, berggrunni, jar�vegi, uppblrestri og afoki, e�a samverkan allra e�a einhverra pessara patta.

Lykilor� : Flokkun i gro�urfelog, reikniaafer�ir moo morgum breytistrerOum (e. multivariate methods), fastir reitir, gr6�ur

breytingar, orfoka land, aburOardreifmg, uppgrreOsla.

Elin Gunnlaugsd6ttir, Institute of Ecological Botany, Uppsala University, Box 559, S-751 22 Uppsala, Sweden.

The Akureyri Museum of Natural History, Box 580, Is-600 Akureyri, Iceland.

(6)

(7)

1. General introduction 1.1 Aims

1. 2 Research plan 1 . 3 Physiography 1 .4 Climate

(a)

as an indication of floristic dynamics 65

7. 7 Discussion ⁶⁵

8. Discussion and conclusion 67

8 . 1 General 67

8.2 Vegetation dynamics 67

Effects of mineral fertilizer 67

8 . 3 Aspects of nature conservation and vegetation development 68

8.4 Evaluation for agricultural use ⁶⁸

8 . 5 Restoration treatments and recommendations for the future ⁶⁸

References 70

Plates 78

(9)

1 General introduction

1 . 1 ^Aims and near the deserted Melakot farm in the deflated

lava-field Eldivi3arhraun, S Iceland (Fig. 1).

The main purpose of the investigation was to study changes in permanent plots in heathland vegetation caused by mineral fertilizer in order to reveal the duration of the fertilizer influence on the vegetation and if the use of the mineral fertilizer will speed up the process of succession. The changes in the soil chemistry caused by the mineral fertilizer were also investigated.

The second main aim was to classify the plant communities in several heathlands involved in the succession studies and to establish the phytosociological status of the vegetation of the permanent plots. Furthermore, attempts were made to relate the plant communities to certain kinds of utilization and to evaluate the agricultural and nature conservation status of the communities judged from their species assemblage.

In the third place it was intended to compare the plant communities distinguished with related com

munities in neighbouring countries and to incor

porate them into the phytosociological system ac

cording to the Braun-Blanquet approach with regard to their climatic and edaphic conditions, as well as their utilization.

1.2 Research plan

The investigations began with vegetation analyses of several plots. Most of them are situated in wind

eroded, deflated areas (i.e., areas where the top soil has been almost totally removed by the wind ero

sion), where the vegetation is under restoration (cf.

Gunnlaugsd6ttir 1982a). The permanent plots were located to three areas of N Iceland, namely the deflated areas of Assandur and Hei3arspor3ur and the dwarf shrub heath in the H6lssandur area. Per

manent plots were also located to several eroded and uneroded areas in the vicinity of Gunnarsholt farm

As the permanent plots of each area are few and the vegetation of these areas had not been described earlier, the investigations were expanded in two of the areas, namely in the deflated Assandur area and in the uneroded heathland in the vicinity of the Gunnarsholt farm. The expanded survey of the ve

getation of these areas facilitated the phytosocio

logical classification and enabled the vegetation in the permanent plots to be assigned to phytosocio

logical units. In addition, the overall survey pro

vides information on the status of the vegetation at the time of the analyses, which is necessary for the long-term successional studies in the areas. In the case of uneroded heathland it also provides an idea of what kind of "original" vegetation is left, a vege

tation where soil is still eroding from its edges.

In the present context the word heathland is used in the sense of Specht (1977 p. 4). The heathland in

cludes dwarf shrub heath, mossy heath (dominated by e.g . , Racomitrium lanuginosum, R. canescensor Hylocomium splendens), grassy heath and open sward with grassy vegetation of deflated areas de

rived from heathland.

Heathlands are wide-spread in Iceland. The dwarf shrub heaths and the open sward heaths are to be found both in the lowland and in the highland.

The mossy heaths, e.g, the Racomitrium heaths, generally belong to the highland vegetation but are also found in the lowlands of SE, S, SW, W and NE Iceland. The grassy heathland is most common in the lowland of S Iceland (cf. Steind6rsson 1 964b).

The data of the vegetation analyses have been processed using multivariate methods. Three data sets are involved, based upon three different methods of vegetation analysis (see below): the first for the analyses of the permanent plots (P-data set), the second for the analyses of the Assandur area (S

data set) and the third for the analyses of the uneroded heathland near Gunnarsholt (H-data set).

(10)

6 EHn Gunnlaugsd6ttir

1 . 3 Physiography

The investigated areas in N Iceland are: Assandur (ea. 20-25 m a.s.l.), Hei3arspor3ur (ea. 360 m a.s.l.) and H6lssandur (ea. 2 1 5 m a.s.l.). There are also several investigated areas in the vicinity of Gunnarsholt farm, in the country of Rangarvellir, S Iceland (Fig. 1) at 100- 1 60 m a.s.l . , namely the lava-field (named Vesturhraun) near Gunn

laugssk6gur, the fenced area near Akurh611, the Rey3arvatnshraum lava-field, the Gari, the Eldivi3arhraun lava-field in Keldnagir3ing near Melakot, all of which are deflated. Also the uneroded heathlands Brekknahei3i, Steinkross

moar, Steinkrossbrun, the heath! and near Kot farm, the Dagver3arnesm6ar heathland and the Su3urhraun and Brejarhraun lava-fields are in

cluded (Fig. 2).

All these areas lie within the zone of volcanic ac

tivities which passes through Iceland. (For the zone see, e.g. , Jakobsson 1 979.) Erosion damage of soils and vegetation is greatest within this zone. The soil there is usually well drained and rather coarse, partly because of the mechanical weathering of the bedrock and partly because of the tephra deposits (and other wind-borne deposits). The coarseness of the soil makes it more sensitive to wind erosion and it retains nutrients poorly. With the exception of Assandur, the areas are lying on basalt lava bedrock of varying age.

The bedrock in the Hei3arspor3ur area is formed of postglacial basaltic and andesitic lavas (Sremundsson 1 977) about 7 1 00 years old (l>orarinsson 1960).

In the vicinity of Gunnarsholt farm there are postglacial basalt la vas (Kjartansson 1 962) from the volcano Hekla. The underlying bedrock in the Gari, the Steinkrossm6ar and the Eldivi3arhraun lava

field is 5500-6000 years old; the bedrock of Steinkrossbrun and Brejarhraun is 5500-6500 years old; the underlying bedrock near Kot farm, the Rey3arvatnshraun and Vesturhraun lava-fields together with the bedrock of the area near Akurholl and the heath Brekknahei3i is 6000-7000 years old, and finally, the extremely irregular and blocky Su3urhraun lava-field is about 1000-5500 years old (Jakobsson 1979).

The Assandur area is unconsolidated gravel flats, formed by the glacial river Jokulsa i Axarfir3i. It

dates back to the finiglacial and postglacial epochs (Sremundsson 1977).

The uneroded 6500-7000 years old lava-fields are covered by a thick loessial, silt-loamy soil with layers of tephra. On the other hand, the young lava

field Su3urhraun has very thin silt-loamy soil in its fissures, or is without any soil. The deflated lava

fields are more or less covered by the same kind of soil, often intermingled with sand and loose blocks.

The bedrock itself commonly rises through the re

mains of the soil.

The soil types of the investigated areas are not widely found in Iceland. Johannesson (1 960) gave a strongly generalized classification of Icelandic soils. He classified the soils of the Assandur area as aeolian sand of flats and dunes, with lag gravel as major associate soil type (his mapping unit 14). This soil type covers ea. 1 .9 OJo of Iceland. Nevertheless, the alluvial gravel flats of the Assandur area fits bet

ter with his mapping unit 16, i.e . , sand, fluvioglacial, a lowland soil type, which covers ea.

2.4 OJo of the total area of Iceland.

The soil of the dwarf shrub heath in H6lssandur and in the uneroded heathlands in the vicinity of Gunnarsholt is silt-loam, 1 5 to 100 cm thick, on gravelly or stony material (his mapping unit 5), a soil type which covers 2.5 OJo of Iceland.

The soil type of the deflated areas, namely Hei3arspor3ur, the Vesturhraun lava-field near Gunnlaugsskogur and the Rey3arvatnshraun, Gari and Eldivi3arhraun lava-fields, is classified as lava where the major associated substrate is aeolian sand (his mapping unit 22), a soil type covering ea 7.9 OJo of Iceland. In my opinion the soil of the above

mentioned areas is better assigned to his mapping unit 1 1 , i.e. , silt loam, 5-15 cm on lava, even if the areas here are deflated. The last-named mapping unit covers 1 . 7 OJo.

Furthermore, it might be of interest to recall that only 27 OJo of the Icelandic soil types are covered by vegetation (J6hannesson 1 960).

1 .4 ^Climate

General characteristics

The climate of Iceland has been classified as boreal in S Iceland and arCtic in N Iceland (Waiter 1979).

Boreal climate (sensu Waiter) has a daily average

(11)

Fig. 1 . Map showing the location of the investigated areas (•) ⁱⁿ Iceland, the meteorological sta

tions ( •) and other places refer

red to in the text.

Heath/and communities in Iceland 7

Fig. 2. The investigated areas near Gunnarsholt farm . The uneroded heathlands are: ( 1 ) Brekknahei5i, (2) Dagverd

arnesm6ar, (3) Steinkrossm6ar, (4) Steinkrossbrun, (5) the level heathland near Kot farm and (6) the Su5urhraun and Brejarhraun lava-fields . The areas of soil deflation are: (7) the Vesturhraun, (8) the Rey5arvathnshraun lava-fields, (9) the Gari and ( 1 0) the Palsteinshraun lava-field.

(12)

8 Elfn Gunnlaugsd6ttir

a b

°C Hella (20ml 50

40 30 20 10

n

d e mm

4.3° 1140 200 ...-1""'"""''--1 00 0

80 60 40 71' 20

k _s

0��0 JFMAMJJASOND

I'

I� �

a �

oc Reykjah 1i8 (285m) mm 2.2°3 94 100

40 80

30 60

20 40

10 20

0 0

q ^r

JFMAM J JASO ND '77

'77 1.5° 442

b

11

'--_....

1974 5.1° 1456

'78 3 .7° 875

� _�

\

1974 2.4° 455

'78 1.8° 357

I

vv

�

'79 1.8° 956

�

'--I-

'75 1.5° 330

'79 0.5° 433

�

'76 1406

'80

'76 2.7°364

·so

(13)

Heath/and communities in Iceland 9

mm 1974 '75 '76

oc Manarbakki (17ml 100 ^{3.9° 632} ^{2.6°4 76} ^{3.9° 418}

3.5°45 1

40 80

30 60

20 40

10 20

0 0

JFMAMJ JASOND

'79

'77 '78 607 '80

2.8° 562 3.0° 416

1.10

c

Fig. 3a-c. Ecological climate diagrams for the stations at Hella, S Iceland; Reykjahlio and Manarbakki, NE Iceland ( 1 93 1 - 1 960). Weather conditions for the period 1 974- 1 980 are also shown. The letters and numbers in the diagrams indicate, (a) station, (b) height above sea level, (d) mean annual temperature CC) , (e) mean annual precipitation (mm) , (k) curve of mean monthly temperature (1 division = woe), ( 1 ) curve of mean monthly precipitation ( 1 division =

20 mm, i . e . , _wo= 20 mm), (m) period of relative drought (dotted) for climate region concerned, (n) corresponding relatively humid season (vertical shading) , (o) mean monthly precipitation _>lOOmm (scale reduced to l!W) ^(black areas , perhumid season) , (q) months with a mean daily minimum below ooe ^(black⁼ cold season), (r) months with absolute minimum below ooc (diagonally shaded) i.e. , with either late or early frosts and (s) number of days with mean temperature above woe.

temperature > + 10°C for less than 120 days and more than 30 days; the cold season lasting longer than 1 80 days. The arctic climate has average daily means > + 1 0°C for less than 30 days and the cold season lasts longer than 240 days.

The climate in S Iceland is better described as oceanic-boreal as the cold period is shorter than 1 80 days because of the influence of the ocean. That type of climate prevails from Hornafjor3ur firth, SE Iceland, along the south coast and over the lowlands of S, SW and W Iceland to the Snrefellsnes peninsula, W Iceland. The oceanic - boreal climate type is also found locally in sheltered places in other parts of the country, but in that case with less precipitation than in the southern part of the country.

In the other parts of the country the climate is oceanic-arctic as the cold period is much shorter than that of the typical arctic climate. The highland climate may be classified as oceanic-alpine-arctic (or subarctic), e.g . , around Grimssta3ir, NE Iceland (see below).

Climate of the studied areas

Climatological data are available from official meteorological stations in the neighbourhood of the studied areas. Reykjahli3 and Grimssta3ir are the nearest stations to the Hei3arspor3ur area and lie 7 km NW and 32 km E of the studied area, respec

tively. The Gar3ur and Grimssta3ir stations lie 1 9 k m WNW and 43 k m SSW o f the studied area in

(14)

10 Eltn Gunnlaugsd6ttir

H6lssandur, respectively. The Garour and Mamir

bakki stations lie 12 km WSW and 27 km NNW of the Assandur area, respectively. The nearest sta

tions to the areas near Gunnarsholt, e.g. , the Vesturhraun lava-field near Gunnlaugssk6gur, are Hella, H�ll and Burfell, which lie 10 km W, 23 km W and 30 km NE of the area, respectively. These sta

tions lie about 23 km WSW, 28 km NW and 25 km NNE of the area near Melakot, respectively (Fig. 1).

Temperature

The means of temperature for three stations are shown in the ecological climate diagrams in Fig. 3, namely for Hella, Manarbakki and Reykjahlio. The figures are based on data from M. A. Einarsson ( 1 976). The means are for the period 1 93 1 - 1 960.

Fig. 3 also shows diagrams drawn in the same way for weather conditions during the period 1 974-Sept. 1 980, i.e . , the period when vegetation analyses of the permanent plots were made, see later. Data are taken from Veonittan ( 1 97 4-Sept.

1 980).

The average annual temperature is highest at the Hella station, S Iceland, and lowest at the Reykja

hlio and Grimsstaoir stations, NE Iceland. July has the highest monthly means at all stations and the lowest ones in January or February.

Interpolated values have been calculated for the studied areas by using constants for temperature changes with growing elevation and distance from the coast given by M. A. Einarsson 1 976. The inter

polated values for Assandur are + 9.8°C in July and -2.4°C in January; corresponding values for Heioarsporour are + 9.9°C and -4. 5°C; for the areas (heathlands) near Gunnarsholt + 1 1 .2 ^oC and -2.4°C; and in the area near Melakot + 1 1 oc and -3 . 1 oc, respectively.

A characteristic of the Icelandic winter weather is frequent thaws, short or long-lasting. The repeated freezing and thawing, and the resulting frost move

ment in the soil, implies a great stress on the plants and their roots.

Precipitation

The average annual precipitation is highest at the stations in S Iceland and is higher at the northern coasts than in northern inland areas. The month of May has the lowest monthly precipitation, with the

exception of one station. The highest monthly records are found sometimes during the period July to October (Fig. 3).

The deflated areas studied here are usually free from snow during the winters, as the snow is im

mediately blown off these unsheltered areas. The same applies to the knolls of the dwarf shrub heath in H6lssandur. The heathland in the vicinity of Gunnarsholt is usually snow-free during winter, or at least with snow cover of short duration.

Means of relative humidity (OJo) for the period 1 958-1 967 are available from the Manarbakki and Hella stations (M . A. Einarsson 1 976), the annual means being 8 1 OJo ^{and 84}^%,respectively.

Evaporation, water deficit

The Assandur, Heioarsporour and H6lssandur areas lie in the part of Iceland with water deficit, especially during the summer months, but also dur

ing the whole year. During the summers the water deficit may be as much as 200 mm in the Assandur and H6lssandur areas, and even more in the Heioarsporour area. All these areas are situated in the rain shadow to the north of the glacier Vatna

jokull. Even the areas in S Iceland have negative values for evaporation during the summer period (see M. A. Einarsson 1 976), which indicates that measurement faults may be involved. However, some of the areas in S Iceland may get positive values.

Wind

Iceland is a windy country like all oceanic islands.

Data for 1 965-197 1 show that SSE winds are most frequent at the Grimsstaoir station, NE Iceland, whereas NE winds are most frequent at the H�ll sta

tion. Calm weather occurs during only 1 9 OJo ^and 24 % of the time at these two stations, respectively.

The average annual wind speed is highest at the coasts and in the highland and the frequency of gales is highest in these parts of the country (M . A.

Einarsson 1 97 6).

It may be pointed out here that southerly winds are relatively dry in N Iceland whereas the northerly winds are relatively dry in S Iceland. The relatively dry winds have contributed greatly to the wind ero

sion damage and the soil deflation. The desiccating and abrasive effects of the winds can clearly be seen in the low, prostrate features of the vegetation and

(15)

Fig . 4. Length of growing season expressed as number of days with

> + 6°C, based on data from M .A. Einarsson ( 1 976) and Bergporsson ( 1 973) .

Fig . 5. Approximate isolines for effective temperature sum, i.e. , sum of day temperatures ex

ceeding ₊6 ° C .

the dried-out soil surface in the open sward vegeta

tion. They are obviously important in preventing plants from becoming established.

Growing season

As the vegetation in the studied areas will be com

pared to related vegetation in neighbouring coun

tries it is of interest as far as possible to compare such important factors for growth and development

Heath/and communities in Iceland 1 1

of vegetation as the length of the growing season and the effective temperature sum in these coun

tries. In the present context, the growing season is considered to be the season with higher temperatures than + 6°C. The growing season in Iceland is thus 60- 160 days long (Fig. 4). Corre

sponding values from Norway are 1 10-200 days (cf. Tu khan en 1 980).

The temperatures of the growing season are low in Iceland. The effective temperature sums (daily

(16)

12 Elin Gunnlaugsd6ttir

E :::l

; 1200

� _<I!

�

Ci ₁₀₀₀

<I>

>

� ·�

w 800

600 400 200

B

• ..

Length of growing season

Fig . 6. Correlation between the effective temperature sum and the length of the growing season (days) , in Iceland and in Norway.

Regression line (A) for Iceland, (B) for Norway. The threshold value for the growing season is + 6°C. Norwegian data are from Tuhkanen (1 980). Icelandic data are from M. A. Einarsson ( 1 976) .

100 120 140 160 180 200 220

degrees) range from 80 to 560 (Fig. 5). Comparable values from Norway are 400- 1 270 (Tukhanen 1 980). The daily degrees are calculated from data given by M. A. Einarsson ( 1 976) by using the for

mula for monthly degrees:

W = (t-t0)

and multiplying the results by 30. In the formula ^W

= temperature sum (heat sum); ^t ⁼ mean temperature for the month and ^t0 ⁼ threshold tern perature.

The relationship between the effective temperature sum (daily degrees) and the length of the growing season is shown in Fig. 6 based on data from Iceland (M. A. Einarsson 1 976) and Norway (Tukhanen 1 980). The Icelandic values are lower in the graph; The equation for the regression line (A) is y ⁼ -353 + 2.80x; ⁿ⁼ 80, ^r⁼ 0.85 and P <

0.001 . The Norwegian values are higher and more to the right in the graph, showing more favourable conditions. The equation for the Norwegian values (B) is y ⁼ -314.5 + 3.65x; ⁿ⁼ 37, ^r⁼ 0.80 and

p < 0.001 .

The effective temperature sum in the Farces is similar to or higher than Icelandic values (Hansen 1 967). The annual, accumulated temperature (a related concept) in the highest parts of the Scottish highlands (Green 1974) is found to be similar to the Icelandic values of effective temperature sum.

1. 5 Vegetation history

General

In these studies attention has been paid to vegetation of deflated areas and to uneroded areas in the neighbourhood of the eroded ones.

Ever since Iceland became settled (874-930) soil erosion has occurred but has become enormous especially during the last three centuries. Indeed, it is impossible to say how much of Iceland (which is 103 1 25 km2) had been covered by vegetation at the beginning of the Settlement era. Some estimate this to about 40 000 km2, of which there are now only 20 000 km2 left. Birch woods and coppices covered about 20 000 km2 at the time of Settlement but at present there are only 1 000 km2 left (I>orsteinsson 1973, I>6rarinsson 1 974).

The Icelandic lowland (i.e., up to an altitude of 350 m) at the time of the Settlement was covered with coppices of Betula pubescens, B. nana and Salix spp. and with birch woods in the most favoured places. Birch coppices were also found in suitable places in the highland up to 450-500 m a.s.l. Today, the highest situated remains of birch are found at 600 m a.s.l. (Steind6rsson 1 964b).

In the fslendingab6k (supposed to be written in the 1 130s), it is stated that the country was covered with woody plants from the coasts to the mountains.

(17)

There are different opinions on how literally these words should be taken (D6rarinsson ^{1 9 74),}but the statement has been confirmed by pollen analyses (D.

Einarsson ^{1 962) .} They indicate that birch dominated at the time of the Settlement but declined soon afterwards; grasses became dominating in

stead.

The far-reaching utilization of the vegetation which was a consequence of the Settlement became, indeed, most hazardous for its future development.

In addition, the deterioration of the climate during 15 50- 1 890 and to some extent volcanic activity had additional damaging influences.

The greatest changes in the Icelandic vegetation is thus the decline of birch woods and coppices, un

doubtedly followed by the successive impoverish

ment of other vegetation types. One or another species might have become extinct after en

vironmental conditions had become too severe (Steind6rsson ^{1 964b) .}The results of these long-term changes are the soil erosion, thickening of the loessial soil (D6rarinsson ^{1 96 1 )}and dune building in some places.

Vegetation history of the studied areas

At the time of Settlement the area now named Assandur was most likely covered by coppices of Salix phylicifolia, S. lanata, Betula pubescens, B.

pubescens x ^{B. nana,}intermingled with herbs like A ngelica spp . , Geranium sylvaticum, Alchemilla spp . , grass and sedge species, by dwarf shrub vegetation, namely vegetation types which still exist in nearby areas . Now there is almost only naked, black gravel and sand left in the Assandur area.

For centuries the Assandur area (then named Asengjar) was a meadow giving good yields (Sigur

j6nsson et al. ^{1 95 8)}which was used for hay-making and grazing (Jar3ab6k Vol. ^{1 1 ,}written in 1 7 1 2) . It became destroyed by debacles in the glacial river Jokulsa i Axarfir3i during the period of 1 694-1729 which were caused by volcanic eruptions at the north edge of the glacier Vatnajokull. The most far-reaching damage to the meadow, where the in

vestigated area lies, occurred in ^{1 694}(D6rarinsson 1 974) . Since then, sand-drift and soil erosion have taken place.

The Hei3arspor3ur area and its vicinity was

vegetated with birch coppices (Sigurj6nsson et al.

1 958) but became gradually wind-eroded, when the coppices became destroyed and the other vegetation overgrazed. Now most of the area is deflated, but remains of older vegetation can be found here and there, and it may be called a dwarf shrub heath with birch coppices. This large area (MY"vatnsorrefi) is a pasture land of Reykjahli3 farm in the county of Myvatnsveit. There is written evidence from ^{1 505} (Dipl. Isl. Vol. ⁷⁾indicating that this pasture was poor in vegetation at that time. In the J ar3ab6k Vol.

1 1 (written in ^{1 7 1 7)}the same area is said to be barren and wind-eroded. During the eruptions in 1 724- 1 729, named Myvatnseldar, the area became covered with tephra (Anmilar 1 400- 1 800, Vol. IV, 6) with sand-drift and soil erosion as a consequence.

The flat land in the surroundings of Gunnarsholt in the county of Rangarvellir was certainly covered by birch wood and willow coppices at the time of the Settlement. Later it became grassy heathland with scattered coppices of birch and willows (Sigur

j6nsson et al. ^{1 958).}Now there are only few places with low shrubs of Salix phylicifolia and S. lanata.

Some species are only found on south-facing slopes in the investigated area, species which un

doubtedly were much more common in the past when the coppices dominated and the microclimate was more favourable, viz. Stellaria graminea, Rubus saxatilis, Prunella vulgaris, A lchemilla vulgaris, Hierochloe odorata and Rumex acetosa.

There are large deflated areas in the surroundings of Gunnarsholt, some of which have been revege

tated during the last five decades and are now used for hay-making and pastures.

The best written document on the situation on the farms in the past is the Jar3ab6k Vol. ²(written in 1 7 1 0), where a description is given of the farms and the utilization ofthe vegetation. The book also men

tions wind erosion and describes a sand-drift, which damaged or ruined the pastures of every farm in the investigated area, e.g. , Kot, Steinkross, Gunnars

holt, Dagver3arnes and Melakot. The two latter farms were deserted at that time. Much land has been destroyed and deflated since ^{1 7 1 2,}especially during the period of cold weather with severe, dry gales (in 1 860- 1 890) . The destruction of vegetation was locally reinforced by overgrazing (sheep) as the yield became severely lowered in the unfavourable weather conditions (Sigurj6nsson et al. ^{1 95 8).}

(18)

2 Material and methods

2.1 Introduction

There are three different data sets as the analyses of the vegetation were carried out by using three differ

ent methods. The results of these methods are com

parable, however, as the same species dominated and the list of recorded taxa was the same.

2.2 Permanent plots In N ^{and S} Iceland (P-data set)

The permanent plots were located in three areas of N Iceland and several in the vicinity of Gunnarsholt (cf. 1 .2 and Gunnlaugsd6ttir 1 982a). Altogether 43 permanent plots were established, 36 of which were subjected to repeated analyses over a varying number of years, all between 1 974 and 1 980. A total of 1 80 analyses were made of the 43 plots. The size (2x6 m2) and the site of the plots were subjectively chosen. Each plot lies on a homogeneous strip of vegetation, each strip being managed in a different way e.g. , with application of mineral fertilizer. At the time of the establishment of the plots, their vegetation was supposed to be "typical" of that kind of management. In each plot percentage cover was estimated for all species and the frequency determined on the basis of recordings in 75 small quadrats of 0.04 m2 size. The plant units were counted in 10 out of the 75 small quadrats. For establishment and results see Gunnlaugsd6ttir (1 982a).

2.3 Line transects at Assandur (S

data set)

The vegetation in the fenced area at .Assandur was analysed by means of line transects, each 4 m in length. Most of them (67) were placed at random and sampled without replacement. A further 12 were situated in the middle of the permanent plots.

The lengths of the transects and their number were arbitrarily chosen, depending on the time available for the investigation.

The line transects were analysed with a perspex frame (see Kershaw 1 964), with pins of 35 or 45 cm length and points of 1 mm diameter. The hits were read at 1 cm intervals, one reading for the first hit in each of the three vegetation layers; the shrub layer, the field layer and the bottom layer ,and in ad

dition, the first hit on litter. The heights of the hits above ground were also recorded (prostrate unmeasurable plant parts were estimated in 0. 1 cm).

Additional species were recorded in the immediate surroundings (2x5 m2) of the line transect.

As the number of hits was high (here 400), and the hits were made with small intervals (1 cm), the results of the readings were transformed to percent

age cover (Kershaw 1 964).

2.4 Systematic plot system for analyses of heathland vegetation near Gunnarsholt

(H

-data set) The vegetation analysis of the uneroded heathland in the vicinity of Gunnarsholt was carried out using a systematic plot system (Fig. 2). The intervals be

tween the plot centres were arbitrarily chosen to be 200 m, and the size of the plots (2x3 m2) satisfies the "minimal area" (cf. Mueller-Dombois &

Ellenberg 1 974) for such communities. For each taxon rooting in the plot, cover was visually estimated in percent by perpendicular projection of biomass.

2.5 Comparison of the sampling methods

The analyses of the permanent plots were very time

consuming as each one included a combination of cover and frequency estimations and the counting

(19)

of plants units (cf. Gunnlaugsd6ttir 1 982a).

By using only cover values, changes of individual species might have been missed (except for the few dominating ones). Only through comparison of plant unit density values can we trace changes in small alpine-arctic species in relation to manage

ment experiments, e.g., the fertilizations. It was necessary to chose the plots arbitrarily as the strips subjected to the different forms of management are small and therefore not convenient for random sampling.

The line transect analyses are also time-consum

ing but the method is more objective. The interpre

tation of the readings by grouping five adjoining figures may overvalue the percentage cover com

pared to the visual estimation of cover, as it was calculated for each of the taxa and then summed up.

Besides, it is more time-consuming to find the ran

domized points in the field than to use the systematic method.

By using the systematic plot system for only cover estimation we save much time. It is possible to make 2-3 times as many analyses in the same period of time as in the line transect method. The latter method is therefore the most convenient for collect

ing overall information for classification of the vegetation in a certain area.

2.6 Soil samples

Sampling

Soil samples for chemical analyses were collected near some of the permanent plots in 1 976 within the strip of vegetation treated in the same way as each one of them. Altogether 21 such samples were col

lected from the 0-5 cm soil layer. Each sample con

sists of 10 subsamples (cf. Gunnlaugsd6ttir 1 982a).

In the uneroded heathland in the vicinity of Gun

narsholt, 48 soil samples were collected for chemical analyses. Most of the sites were sampled at random.

Each of the soil samples consists of 20 subsamples taken from the 0-5 cm soil layer in the same kind of vegetation as found in the analysed plots.

Chemical analyses

For the analyses of soil samples collected in 1 976, see Gunnlaugsd6ttir ( 1 982a).

The soil samples collected in the heathland i n 1 982 were dried at 25°C. The soil samples were sieved to obtain a particle fraction of< 2 mm. The pH was measured in H20. The solvent for available phosphorus was 0.5 N NaHC03, adjusted at pH 8 . 5 ; the samples were shaken for 30 min (Olsen et al., 1 954). The solvent for the potassium analyses was 1 .25 o/o acetic acid; the samples were shaken for 30 min.

2. 7 Data analysis

Introduction

The (estimated) cover-abundance data of the vegetation analyses have been transformed into the ordinal scale (van der Maarel 1 979) for multivariate treatments.

Readings all along the line transect were lumped together in groups of five adjacent ones (cf. Ker

shaw 1964) and transformed to percentage cover.

The results of the classification of the vegetation are based on these cover percentages; additional taxa have been evaluated as I in the ordinal transforma

tion scale.

Multivariate treatments

Multivariate methods have been used both for classification and subsequent typology of vegeta

tion and for ordination, i.e. the arrangement of phytosociological entities along one- or multi-di

mensional schemes according to the relation be

tween the entities (see van der Maarel, 1 979 and references quoted there).

Two data programs have been used in these par

ticular studies for the classification of vegetation, namely the T ABORD program for the classification and the ORDINA program for the ordination.

a) The TABORD program (van der Maarel et al.

1 978) is a procedure for clustering releves based on their similarity in combination with a procedure for obtaining a diagonal structure of clusters in a table.

The steps of the program are: Establishment of ini

tial clusters, relocation using the similarity ratio (Wishart formula), homogenization of clusters, construction of a table, inspection of the results in terms of optimal number of groups and their

(20)

1 6 Elfn Gunnlaugsd6ttir

homogeneity and repetition of the relocation and fusion procedure with other options.

The following options were used throughout the treatments: The minimum cluster size was chosen as 1 because there are several releves of special floristic composition involved. The threshold value was chosen as 0 to avoid residual groups, and the fusion level as 0.5 to get clusters homogeneous enough for interpretation as associations (cf. Westhoff & van der Maarel 1 978). A high fusion level would result in too many, though more homogeneous, clusters at a lower phytosociological level. The frequency limit was chosen as 0.6 to obtain fairly constant species in the blocks of differentiating species.

b) The ORDINA program (Roskam 1 971) per

forms a centred, non-standardized principal com

ponent analysis (PCA), based on a Euclidian distance matrix, following Orloci (1966). The percentage of explained variances are calculated, and a measure of success relating the new distances to the original one is given.

The minimum number of clusters chosen varies from one data set to another. In the P-data set the minimum number is 1 5 ; in the H-data set it is 16 and in the S-data set 12 or as many as the expected number of community types in each set. The initial cluster array of the P-data set was obtained with the systematic-random device provided by the pro

gram. Cluster arrays for the other two data sets were devised on the basis of field knowledge.

(21)

3 Statistics on species occurrences

3 . 1 Introduction

Altogether there are 203 taxa involved in the various data sets; 1 52 in the H-data set, 73 in the S-data set and 1 58 in the P-data set.

The taxa of vascular plants amount to 105 , belonging t o 3 2 different families (see Table 1). The total number of cryptogam taxa (lichens, mosses, liverworts) is 98 (see Table 2 for details).

3 . 2 Relation between number of taxa and number of plots

The number of plots in relation to the number of species per plot for each of the two data sets (H and S) is shown in Figs. 7a & 7b. Clearly the deflated areas are much poorer in species. The mean numbers of taxa per plot are 25 and 12 respectively.

The distribution curve for the uneroded plots is almost normal, that of the deflated plots is skewed to the right (Fig.7b). The distribution curve for the P-data set (Fig. 7c) is still more skewed because of the more pronounced heterogeneity of the data.

However, the mean number of taxa per plot is 23 , almost as high as in the uneroded plots. The same relation for the first analyses of each of the 43 per

manent plots shows still more discontinuity (Fig.

7d). The P-data set is mainly from deflated areas and also from uneroded areas as well as from species-rich revegetated areas, which contributes to keep the mean number of taxa per plot relatively high.

3 . 3 Relation between cumulative number of taxa and number of plots The cumulative number of species was determined in relation to the cumulative number of plots in order to investigate the species-area relation in the

entire data set. The present 79 samples were taken of the S-data set with 73 taxa (Fig. 8a) and 201 samples of the H-data set with 1 52 taxa (Fig. 8b).

In addition there are 43 samples, i.e., the first analyses of each of the permanent plots from the P

data set with 1 30 taxa (Fig. 8c).

Clearly the number of species increased with the total number of samples involved and no plateau is reached. This suggests that the species-area relation is a logarithmic one, either according to Williams a semilogarithmic (Dahl 1 956, Williams 1 964), or a double-logarithmic one according to Preston (1962). In our case the Preston curve gave best fit.

For the 43 plots of the P-data set and 1 30 taxa the equation is log y ⁼ 1 . 1 + 0.66 log x from which it was calculated that about 35 plots are sufficient to get complete taxa composition.

The equation for the S-data set of 79 samples and 73 taxa is log y ⁼ 1 . 12 + 0.39 log x from which it was calculated that about 80 plots are sufficient to get complete taxa composition.

For the H-data set of 201 samples and 1 52 taxa the equation is log y ⁼ 1 .62 + 0.24 log x, which reveals that about 2 1 5 plots are sufficient for com

plete taxa composition.

3 .4 Phytogeographical observations The geographical distribution of the species in

volved in these studies, both in Iceland itself and in the neighbouring countries, will be treated briefly below.

Many of the species are mountain species in the neighbouring countries, although they grow in the lowland in Iceland. About half of the species studied are Arctic species and the other half European ones (cf. classification by M0lholm Hansen 1 930), the European species mainly being found in the north

ern groups (Table 3). The Arctic species are species of common occurrence in the arctic and subarctic regions, but they are absent or rare in temperate

(22)

1 8 Elln Gunnlaugsd6ttir

Table 1 . N umber o f vascular p lant taxa per fami l y .

Data set : H s p All

Fam i l i es :

1 Poaceae 1 5 1 1 1 8 20 2 Caryophyllaceae 4 8 1 0 1 0

3 Cyperaceae 8 2 3 8

4 Asteraceae 5 1 4 5

5 Polygonaceae 2 1 5

6 Gentianaceae 4 5

7 Brassicaceae 2

8 J uncaceae 3

9 Salicaceae 3

1 0 Rosaceae 4

1 1 Equisetaceae 3

1 2 E ricaceae 3

1 3 R ubiaceae 3

1 4 Saxifragaceae 2

1 5 Betulaceae 1

1 6 Ophioglossaceae 1

1 7 Empetraceae 1

1 8 Ranunculaceae 2

1 9 Scrophulariaceae 2 2 0 Lam^iaceae

21 O rchidaceae

2 2 Plumbaginaceae

2 3 Liliaceae 2 4 Plantaginaceae

2 5 Lentibulariaceae

2 6 Crassulaceae 2 7 Selagi nel laceae 2 8 V iolaceae

2 9 Athyriaceae

30 Fabaceae

3 1 Onagraceae

3 2 Pyrolaceae

Total numbe r : 7 9 4 9 8 8 1 04

Number 15 of plots

10

5

5 10 15 20 25

Number 20 of plots

15

10

5

5 10 15 25 30

-r-·

LJ �

of plots

5 10 15 20 J:l 25

30

Table 2 . N umber of taxa belonging to different cryptogam groups .

Data set : H s p All

Lichens

folio se 9 3 8 1 1

fructicose 2 1 1 0 1 6 2 7

crustaceous I I 1

B ryophyta

liverworths 1 4 1 3 1 7 mosses

acrocarpeous 1 9 2 3 30

p leurocarpeous 8 8 I 0

cladocarpeous 2 2 2

Total numbe r : 7 3 2 4 7 1 9 8

Table 3 . Percentage o f Arctic a n d European taxa (or species sensu Molholm Hansen 1 930} w ithin the data set s .

Data set : H s p _A_II ^_^_

Arctic taxa ( 45. 7}

A 1 1 2 . 7

A 2 1 2 . 7

A 3 2 0 . 3

E uropean taxa ( 5 1 . 9)

E 1 6. 3

E 2 1 1 . 4

E 3 1 6. 5

E 4 1 7 . 7

Ungrouped taxa 2. 4 1 00 . 0

a

35 40 45

c

35 40 45

d

( 48. 9) 1 0 . 2 1 6 . 3 2 2 . 4 ( 4 5 . 0}

4 . 1 8. 2 1 8 . 4 1 4 . 3 6. 1 1 00. 0

50 55

taxa/plot

( 44 . 9 ) ( 4 5 . 7) 1 0 . 2 8. 6 1 4 . 9 1 7 . 1 1 9 . 8 2 0 . 0 ( 5 1 . 7 ) ( 49. 5 )

4 . 6 5. 7 1 3 . 8 1 0 . 5 2 0 . 7 1 9. 0 1 2 . 6 1 4 . 3 3 . 4 4. 8 1 00 . 0 1 0 0 . 0

Number 10 ofplots

5

5 10 15 20

r:l 0 I I 0 0 I I r:l r:l

Fig. 7 . The frequency distribution of taxa number/plot. (a) Data from the H -data set . (b) Data from the S-data set . (c) Data from the P-data set. (d) Data from the first year of analysis of the 43 per

manent plots. Mean number of taxa is ea. 25 , 12, 23 and 23 , respectively.

30 35 40 45 50

(23)

2.0

Log no.of taxa

1.6 0

• •

1.2 •

0.8

0.4

0.4 0.8 1.2 1.6

Log no. of plots 2 .0 2�

zones. The European species are common or typical of more southerly regions. Further subdivsion of both groups into minor groups (the Arctic species into three minor groups, Al-A3 and the European species into four minor groups El -E4) is according to the northern limits of the species. The species with the highest values extend farthest north in each of the groups (cf. M0lholm Hansen 1 930 pp. 20 ff.) .

Fig. 8 . The logarithmic relationship between the cumulative number of taxa and the number of plots for (a) 201 plots in the heathlands near Gunnarsholt farm, data from the H-data set, (b) 79 plots in the Assandur area, data from the S-data set and (c) the 43 permanent plots , a part of the P-data set.

Composition and dynamical status of heathland communities in Iceland in relation to recovery measures

Composition and dynamical status of heathland communities in Iceland

in relation to recovery measures

Composition and dynamical status of heathland communities in Iceland

in relation to recovery measures

Contents

(a)

1 General introduction

I� �

11

� �

�

�

�

�

2 Material and methods

(H

3 Statistics on species occurrences

-r-·

LJ �

� _�