Vegetation and soils of seminatural deciduous woods in northeast Ireland*

Alan Cooper** Department of Environmental Studies, University of Ulster (Jordanstown), Shore Road, Newtownabbey, County Antrim, BT37 0QB, Northern Ireland

Keywords: Basalt, Conservation, Deciduous wood, DECORANA, PCA, TWINSPAN, Woodland soil

Abstract

Seminatural deciduous woods associated mainly with basaltic escarpments were surveyed in northeast Ire- land to investigate the relationships between soil physical and chemical characteristics and community com- position. Gradients of floristic composition in the sample stands pointed to soil acidity, soil mineral nutrient status and the degree of soil waterlogging as principal factors influencing the ground flora.

The physico-chemical characteristics of the woodland soils showed distinctive differences which were cor- related with community composition, confirming the considerable effect of the soil environment. Soil texture was shown to be a key factor influencing the ground flora, mainly through its influence on soil water status and acidity.

Better drained soils subject to minimal disturbance by people or physiographic processes, were found to have the highest content of water soluble nitrogen, available phosphorus and exchangeable potassium. An interaction between soils, site geomorphology and land use suggested that the value of the woods as controls against which the effects of agricultural land use on soils can be monitored, is limited.

Introduction

Seminatural woods dominated by Corylus avel- lana (hazel) are commonly associated with escarp- ments of Tertiary basalt in northeast Ireland. Many of the woods are probably ancient, albeit managed relicts of primary woodland sensu Rackham (1981), if assessed on the basis of their content of primary woodland indicator species (Peterken, 1974) but there are few archives to prove this. Large trees are infrequent and the open canopy promotes a high species diversity in the ground flora (Cooper, 1984a). Aspects of the ecology and conservation

* Nomenclature follows Webb (1977) for vascular plants and Watson (1968) for bryophytes. ** A grant from the British Ecological Society helped to finance part of the field work for this project. Gary McClure carried out soil physico-chemical analysis.

Vegetatio 64, 67-74, 1985. © Dr W. Junk Publishers, Dordrecht - Printed in the Netherlands.

management of hazel woods in northeast Ireland have been discussed by Cooper (1984b).

The phytosociology of southern and western Irish woods has recently been updated (Kelly & Moore, 1974; Kelly, 1975; Kelly & Kirby, 1982), but there are few data on the variation of associated soil physico-chemical characteristics. Ball & Stevens (1980) discussed the conservation value of woodland sites where soils have not been modified repeatedly by human influence. Such sites, where pedogenesis under woodland conditions has proceeded for at least several hundreds of years, are particularly valuable as controls for the compara- tive study of soil development and the effects of people on their environment.

The objectives of the present study were to inves- tigate the range of variation in seminatural wood- land vegetation and soils in northeast Ireland and to comment on the relationship between soil physi-

68

cal and chemical characteristics and community composition.

Methods

A study area of ca 40 × 40 km in northeast Ireland just north of Belfast was selected to include representative landscape categories present in the region. Using the Irish Grid 1:10000 O.S. maps, woods shown as hardwood stands were identified. Of those still extant, 22 sample sites were selected at random. Woods were not considered for field survey if there was evidence that they had been planted with introduced trees or if they did not con- tain predominantly indicator species of primary woodland (sensu Peterken, 1974).

Within the woods, 53 sample stands were dis- tributed randomly, one to each habitat in a wood. Habitats were identified subjectively on the basis of sudden changes in topography associated with, for example, stabilised scree slopes, rock piles, stream sides and flushes, to stratify the samples. Field work was carried out in July and August of 1981, according to the technique developed for woodland survey by Bunce & Shaw (1973) but the largest quadrat size used was 100 m 2 and only qualitative records were taken for bryophytes.

The two-way indicator species analysis TWIN- SPAN (Hill, 1979a) of qualitative records for all vascular plants and bryophytes in the sample stands and the detrended correspondence analysis DECORANA (Hill, 1979b) were used to classify and ordinate the field data and the program pack- age DATAEDIT (Singer, 1980) facilitated the preparation of summary tables.

Groups of quadrats generated by TWINSPAN were used to stratify a soil sampling programme. Three sample quadrats were chosen subjectively from each group. The samples represented typical examples of the woods, in terms of their floristic composition and site characteristics. This approach was taken to reduce sample variability and increase the cost effectiveness of the soil analysis.

At each sample quadrat in the field, 10 random samples of the surface 12.5 cm of the soil profile were taken after removing sur- face litter, using a 2.5 cm diameter~stainless steel auger. Most of the soils were thin and rocky so that where soil depth was less than 12.5 cm, effectively the whole of the profile was sampled. Sample randomisation was achieved with a table of random

numbers and a magnetic compass. All soil samples were collect- ed on the same day, 5 days after the last recorded rainfall. Sam- ples were transferred immediately to an airtight container where they were bulked as collected. On returning to the laboratory, samples were weighed fresh and pH determinations were carried out. Soils were air-dried for three weeks before further analysis.

Standard methods of soil analysis given by Allen et aL (1974) were used to determine soil physico-chemical characteristics. Water content was determined gravimetrically on air-dried soils. After this, each sample was lightly pulverised in a pestle and mortar, then passed through a 2 mm brass sieve to determine the small stone fraction. Particle size analysis was carried out on the resulting fine earth using the hydrometer method. Water-soluble nitrate concentrations were estimated polaro-graphicaIly. Avail- able phosphorus was extracted with Truogs reagent and ex- changeable cations were extracted with a 2.5°7o v/v solution of acetic acid. Organic carbon was determined by the Walkley and Black (1934) method.

Stand elevation was recorded from 1:10 000 O.S. maps and the presence of scree or rock outcrops and evidence of grazing ac- tivity were recorded in the field. Also estimated at each sample quadrat were nominal indices of slope (0- 3:flat-very steep) and the abundance of large (>50cm) boulders (0-3:absent- abundant).

The matrix of soil physico-chemical attributes was standardised as a percentage of the maximum value of each attribute and a centred principal components analysis (PCA) ordination was carried out on this data, using the program suite OR- DIFLEX, (Gauch, 1977). Correlation coefficients were calculated between stand scores on the first three axes of this ordination and stand scores on the first three axes of the DECORANA ordination based on stand floristic composition.

Results

The first division of the TWINSPAN classifica- tion contrasted quadrats containing species charac- teristic of acidic soils with those in which species from more base-rich soils were common (Webb, 1977; Ellenberg, 1974). Indicator species (sensu Hill, 1979a) are listed in Table 1. A second division isolated a group of quadrats containing species usually found on well drained woodland soils from quadrats with species more tolerant of soil water- logging (Table 1). The distribution of the three ma- jor groups of quadrats on the DECORANA ordi- nation is shown in Fig. 1. The first axis of floristic variation was interpreted as representing a gradient of decreasing soil acidity and nutrient status. The second axis represented a gradient of decreasing soil water content.

Table 1. Species composition of the three main woodland com- munities delimited by a TWINSPAN classification. 1. Species occurring in a significantly greater number of quadrats than ex- pected from a random distribution in the whole sample set. Sig- nificance (p=0.001) was determined using the Fischer exact probability test (Sokal & Rohlf, 1981). 2. Indicator species (sen- su Hill, 1979a) given by the TWINSPAN classification but not showing any significant group preference,

Well drained; nutrient-rich Hea~, a helix 1, 2

Poorly drained, nutrient-rich Veronica officinalis 1 Allium ursinum 1 Filipendula ulmaria 1, 2 Veronica montana 1 Deschampsia caespitosa 1 Lysimachia nemorum 2 Cireaea lutetiana 2 Rhytidiadelphus triquetrus 2

Acidic Dicranum majus 1, 2 Mnium hornum 1, 2 Holcus lanatus 1, 2 Blechnum spicant 2

Quadrats with an acidic ground flora were found at higher elevations, on sites which were moderately steep to steeply sloping, often with heavy or organ- ic soils and with large boulders occasional to abun- dant. Quadrats from the base-rich, better drained

2~o 1 +

175-

O

O ©

u3

_o

2 3o C

o _o 3

O t3

0 _o

O A • • •A•

AXIS 1 175 250

Fig. I. A DECORANA ordination of the woodland quadrats showing the distribution of the three main groups identified by a TWINSPAN classification: nutrient-rich, well drained woods (o ) ; nutrient-rich, poorly drained woods ( • ) ; high elevation acidic woods ( [] ). Quadrats from which soil samples were taken have been underlined.

69

stands were from lower elevations and were found on either steep to very steep stabilised scree or soils with abundant large boulders. The third group of poorly drained quadrats were also from lower ele- vations and some were associated with the moder- ately steep slopes of stream sides and flushed sites.

The first three axes of the PCA ordination of soil samples accounted for 36.3%, 19.9% and 12.7% of the total variability respectively (Fig. 2). Soil characteristics associated with differences in soil texture were contrasted along the first axis, with

(a)

100 • S O D I U M

I I b H 80

•CALCIUM •MAGNESIUM

6O

4O

2O • STON E S

SAND O

20

• POTASSIUM

•NITROGEN

°PHOSPHORUS

•BOULDERS

• WATE R

•SILT

i i i

40 60 80 100

A X I S 1

• ORGANIC CARBON •CLAY

(b)

100

80

60 cO

co

40 • MAGNESIUM

•STONES •CALCIUM

•SODIUM bSAND

•pH

•PHOSPHORUS

•BOULDERS

• NLTROGEN

• POTASSIUM

WATERoosIL T

i I i J i

20 40 60 80 100

A X I S 1

o ORGANIC CARBON

•CLAY

Fig. 2a+b. The first three axes of a centred PCA ordination of soil physico-chemical characteristics; (a) axes 1 and 2, (b) axes 1 and 3.

70

Table 2. Correlation coefficients between stand scores on the first three axes of a PCA ordination of soil physico-chemical at- tributes and stand scores on the first three axes of a D E C O R A N A ordination of floristic data. Significant correla- tions (p= 0.05) have been underlined.

PCA D E C O R A N A

axis axis axis axis axis 2 3 1 2 3

PCA axis 1 -0.006 -0.059 0.554 -0.348 0.113 PCA axis 2 -0.004 0.082 -0.528 -0.043 PCA axis 3 -0.229 -0.450 0.208 D E C O R A N A

axis 1 -0.128 0.226 D E C O R A N A

axis 2 0.126

Table 3. Percentage frequency (a) and mean percentage covei (b) of species present in the six groups generated by a TWIN- SPAN classification of woodland quadrats . Only species with a frequency of > 80O7o or a mean cover value of 10o70 or more in any one group of quadrats have been tabulated. Frequency values are given in deciles ( + and 1 - 9). Communit ies and spe- cies are arranged according to their loadings on a reciprocal averaging ordination (Singer, 1980).

T W I N S P A N Group Number 1 2 3 4 5 6

a) Holeus lanatus - 9 I + -

Luzula sylvatlea 8 4 - + 2 Mnium hornum 7 9 4 2 2 Blechnum spicant 7 8 1 5 2 Fragaria vesca 9 8 4 5 5 2 Sanicula europaea 9 5 7 7 4 2 Sorbus aucuparia 7 9 - 1 4 4 Hedera helix 8 + 7 2 7 2 Plagiochila asplenioides 9 9 5 9 5 4 Rubus fruticosus 8 8 9 8 9 4 Pellia epiphylla 8 1 - + 6 Corylus avellana 9 9 8 9 8 6 Arrhenatherum elatius 8 5 4 5 4 6 Hypnum cupressiforme 7 9 7 9 5 6 lsothecium myosuroides 8 2 2 5 2 Brachypodium sylvaticum 8 5 5 4 4 8 Geum urbanum 7 7 8 8 5 8 Oxalis acetosella 7 9 5 9 5 8 Primula vulgar& 8 5 7 7 5 8 Viola riviniana 7 9 7 9 8 8 Eurynchium praelongum 9 8 9 9 8 9 Lophocolea cuspidata 8 8 8 9 5 9 Dryopteris dilatata 6 5 7 2 5 8 Crataegus monogyna 5 7 7 7 5 8 Athyrium filix-femina 9 8 5 9 7 9 Fissidens taxifolius 7 7 5 8 5 8 Eurynchium striatum 7 7 8 8 5 9 Thuidium tamariscinum 8 8 5 7 8 9 Hyacinthoides non-seriptus 1 7 5 9 2 6

Table 3. (Continued).

TWINSPAN Group Number 1 2 3 4 5 6

Potentilla sterilis 5 6 8 6 5 9 Geranium robert&hum 2 5 4 6 7 8 Epilobium montanum 3 4 5 4 7 8 Arum maculatum 2 1 1 8 2 Poa trivialis 1 9 2 5 9 9 Mnium undulatum 2 5 5 9 5 9 Lysimachia nemorum 3 7 9 5 9 Fraxinus excelsior 2 3 2 5 7 8 Circaea lutetiana 2 4 - 5 - 8 Ajuga reptans 2 4 5 2 8 Ranunculus repens 2 2 1 - 9 Filipendula ulmaria 1 - 1 7 4 9 Rumex sanguineus - + - 1 5 8 Allium ursinum - 5 1 8

b) Luzula sylvatica 20.0 0.2 11.0 0.i - Hederahelix 3.6 1.0 0.l 0.9 7.0 10.0 Brachypodium sylvaticum 21.0 14.0 0.6 11.0 3.2 1.0 Oxalis acetosella 3.9 1.6 13.0 8.0 16.0 3.0 Athyriumfil ix-femina 18.0 16.0 4.8 8.0 9.0 6.0 Circaea lutetiana 1.4 10.0 0.4 2.4 - Dryopterisfilix-femina 7.0 1.4 1.3 0.6 14.0 0.4 Heracleum sphondylium 3.9 2.2 0.1 0.7 12.0 5.0

Number of species 115 147 80 132 96 107 Number of samples 8 12 7 14 7 5

coarse, dry, base-rich soils separated from heavy, poorly drained, more acidic soils. The second axis separated soils according to their exchangeable ca- tion status. Well drained, base-poor soils had lower scores on this axis than well drained, base-rich soils. Soils with a high water content also had high scores on the second axis. The third axis separated soils with higher concentrations of nitrogen, phos- phorus and potassium. These soils were well drained, had a high organic matter content and large boulders were abundant.

There were significant correlations between stand scores on the axes 1 and 2 of the PCA soil and the DECORANA vegetation ordination, and a significant correlation between stand scores on axis 3 of the soil ordination and axis 2 of the vegetation ordination (Table 2). These results point to a strong relationship between community composition and the edaphic environment and suggest that soil tex- ture, mainly through its influence on soil water sta- tus and acidity, is a key factor influencing the ground flora.

71

Table 4. Mean values of soil physico-chemical characteristics for the groups generated by a TWINSPAN classification of sample quad- rats. Any two means are significantly different if they differ by an amount greater than the least significant range (Parker, 1979).

TWlNSPAN Group Number

1 2 3 4 5 6

Least significant range

Clay fraction (% dry wt. of fine earth) 9.7 18.6 9.3 8.9 15.6 12.5 n.s. Silt fraction (°70 dry wt. of f.e.) 31.0 49.1 27.9 42.4 37.4 44.9 15.7 Sand fraction (% dry wt. of f.e.) 51.2 21.2 58.1 41.3 34.7 41.0 26.5 Stone fraction (% dry wt.) 15.6 14.8 32.0 15.7 13.7 23.0 15.9 Water content (% dry wt.) 35.3 43.7 13.3 33.7 36.2 46.6 15.7 Organic carbon content (07o dry wt. of f.e.) 4.6 3.8 2.8 4.3 6.9 5.8 3.2 pH 5.4 4.9 5.0 5.4 4.8 5.3 0.62 Exchangeable calcium (meq 100g-~) 26.9 13.3 21.8 22.6 13.7 29.8 16.7 Exchangeable magnesium (meq 100g ~) 15.0 6.5 15.9 10.0 6.8 13.0 n.s. Exchangeable potassium (meq 100g -~) 1.68 1.47 1.17 1.89 1.45 1.27 n.s. Available phosphorus (mg 100g ~) 2.59 1.52 1.99 1.36 2.05 1.02 n.s. Water soluble nitrate (mg 100 g- ~) 0.17 0.48 0.41 0.25 0.62 0.31 n.s.

Further TWINSPAN divisions of each commu- nity could be interpreted in terms of the contrasting ecology of their component species up to a total of six quadrat groups. Group nomenclature is based firstly on the most frequent species and secondly on the most abundant species in each group. Table 3 summarises group composition and their soil physico-chemical characteristics and other environ- mental attributes are given in Tables 4-5 .

The Sanicula europaea/Luzula sylvatica vegeta- tion (group 1) contains freely drained soils often on steeply sloping rocky escarpments. These soils are moist and organic and have a relatively coarse, sandy mineral matrix and high concentrations of exchangeable cations. Many of the preferential spe- cies identified with TWINSPAN are characteristic of acidic brown earth soils occurring in ungrazed woods with a humid microclimate: for example

Table 5. Site attributes estimated for each of the six main groups generated by a TWINSPAN classification of woodland quadrats.

TWINSPAN Group Number

1 2 3 4 5 6

Mean stand elevation (standard deviation)

Frequency of stands in classes of slope (0-4: flat-very steep)

Number of stands with scree or rock present

Frequency of stands in classes of boulder abundance (0 - 3: absent-abundant)

(In) 180 140 157 160 117 203 (55) (17) (40) (35) (58) (25)

0 1 1 1 1 2 1 2 2 1 2 1 2 3 1 1 1 4 1

1 1 3 1 3

0 3 3 1 1 3 1 1 2 2 1

3 2 1

72

Luzula sylvatica, Pellia epiphylla, Asperula odora- ta, Blechnum spicant and Hookeria lucens.

The soils of the Holcus lanatus/Circaea lutetiana vegetation (group 2) have high silt and clay frac- tions and are poorly drained. They usually overlay moderately steeply sloping glacial drift. There is a high soil water content and concentrations of ex- changeable cations are low. Samples are charac- terised by preferentials (sensu Hill, 1979a) such as Conopodium majus, Carex sylvatica and Holcus mollis. These species are generally found on nutrient-poor, heavy, clay soils.

The Potentilla sterilis/Oxalis acetosella vegeta- tion (group 3) occur on soils which are freely drained and overlay steeply sloping stabilised scree. There is a high sand and stone fraction and a low water content. Concentrations of soil exchangeable cations are high but the organic carbon content is low. This group contains preferential species such as Teucrium scorodonia, Stellaria holostea and An- thoxanthum odoratum which are generally found in drier, more open, ungrazed woods.

In the Hyacinthoides/Brachypodium sylvaticum vegetation (group 4), the soils are often associated with stream sides. The woods are usually found in sites at lower elevations. There is a high silt and sand content and high concentrations of exchange- able cations. Included amongst preferentials are Deschampsia eaespitosa, Veronica officinalis, Prunella vulgaris and Sanicula europaea, species of heavy but not permanently waterlogged soils in Ire- land.

Soils of the Arum maculatum/Dryopteris filix- femina vegetation (group 5) are freely drained and have the highest organic carbon contents. They usually occur at the bottom of scree slopes and have a high clay fraction but low concentrations of soil exchangeable cations. The soils are rich in water-soluble nitrogen in comparison with the oth- er soils and the concentration of available phos- phorus is also high. Preferential species are Arum maculatum, Rumex sanguineus and Urtica dioica, indicators of nutrient-rich, well drained soils of un- grazed woods. Hedera helix, also a species charac- teristic of ungrazed woods in Ireland, is a group dominant.

The Filipendula ulmaria/Hedera helix vegetation (group 6) was flushed with laterally moving ground water and has a high organic carbon content. There is a high silt and sand fraction and high concentra-

tions of exchangeable cations. Species such as Jun- cus effusus, Valeriana officinalis, Chrysosplenium oppositifolium, Alnus glutinosa, Acrocladium cus- pidatum and Equisetum telmeteia, which are usual- ly found in habitats where soil waterlogging is fre- quent, are preferential. Species with a high frequency included Filipendula ulmaria, also characteristic of waterlogged soils, and several plants able to tolerate grazing, namely Poa trivialis and Ranunculus repens.

Discussion

The three main communities generated by the TWINSPAN classification correspond broadly to phytosociological associations already described in Ireland. The acidic woods are similar to the Blechno-Quercetum coryletosum (Kelly & Moore, 1974). A preferential species given for this subas- sociation but absent from the woods in the present study was Euonymus europaeus. Several other differential and character species of higher group- ings were not found. The better drained, more base- rich stands are related to the Corylo-Fraxinetum typicum (Kelly & Kirby, 1982) but the absence of some character- and differential species is probably caused by the difference in parent materials (basalt versus limestone) from which the two sets of sam- ples were taken.

Some of the less poorly drained stands resemble the Corylo-Fraxinetum veronicetosum but most of the wetter sites are more similar in species composi- tion to the Alno-Padion. A recent classification of British woods by Bunce (1982) indicated that many of the woods have a ground flora that is similar to Quercus petraea-Fraxinus woodlands of the Dryoptero-Fraxinetum (K16tzli, 1970) but do not have a strong oceanic character. The woods resem- ble those described on exposed Tertiary basalts in western Scotland (Birks, 1973; McVean & Ratcliffe, 1962; Jermy & Crabbe, 1978).

The stratification of the woodland quadrats on the basis of their floristic composition proved a convenient and efficient method of sampling the edaphic variation in the hazel woods studied, judg- ing by the ease with which the PCA ordination of soils was interpreted. Even with only three replicat- ed soil samples from each TWINSPAN group, sig- nificant differences in texture, nutrient status, acidity and water content were evident.

Differences in soil texture, acidity and water sta- tus related to differences in site geomorphology and elevation are the key factors governing the spe- cies composition of the woods. Also, people have had a great influence on their structure and composi- tion mainly through the selective felling of trees and the secondary use of woods as shelter for graz- ing animals (Cooper, 1984a). Stands with the highest loading on the third axis of the PCA ordi- nation of soil attributes and therefore with the highest concentrations of nitrogen, phosphorus and potassium, were the slightly acidic, well drained but organic soils often at the bottom of stabilised scree slopes. These were the least dis- turbed woodland habitats in terms of geomorpho- logical processes and human impact. Also with high loadings on the third axis were soils from stands on the less steeply stabilised scree slopes and the better drained stands within woods on generally poorly drained substrates. Stands with the lowest loadings on the third axis were either waterlogged, or from higher elevation, acidic soils, or from less stable scree slopes.

Cooper (1984a) summarised the conservation value of hazel scrub communities in northeast Ire- land: woods associated with stabilised scree slopes, and with boulders at the base of cliffs and scree slopes, are effective barriers against domestic graz- ing animals and changes in land use brought about by woodland clearance for agriculture. The natural- ness of the soils on which these woodlands are found adds to the conservation value of the sites.

It could be argued that soils associated with cliff bottoms and stabilised scree slopes probably have the greatest intrinsic scientific interest because they may represent woodland soils that once covered large areas now occupied by agricultural land. McConaghy & McAleese (1957) and Brown (1951, 1954) described the physico-chemical characteris- tics of agricultural soils derived from basaltic gla- cial drift in northeast Ireland. Even the most finely textured soils of the present study, however, had a clay content less than half that of the coarsest agricultural soils. The interaction between soil physico-chemical characteristics, site geomorphol- ogy and land use, therefore, suggests that at least for northeast Ireland, the value of semi-natural woods as controls against which the effects of agricultural land use can be measured, is limited.

73

References

Allen, S. E., Grimshaw, H. M., Parkinson, J. A., Quarmby, C. & Roberts, J. D., 1974. Chemical Analysis. In: S. B. Chap- man (ed.). Methods in Plant Ecology, p. 411-466. Black- well, Oxford.

Ball, D. F. & Stevens, P. A., 1980. The role of 'ancient' wood- lands in conserving 'undisturbed' soils in Britain. Biol. Con- serv. 19: 163-176.

Birks, H. J. B., 1973. Past and Present Vegetation of the Isle of Skye: a Palaeoecological Study. Cambridge University Press, Cambridge, 415 pp.

Brown, W. O., 1951. The cation exchange properties of the basaltic drift soils of north-eastern Ireland. Emp. J. Exper. Agric. 19: 134-138.

Brown, W. O., 1954. Some soft formations of the basaltic region of North-East Ireland. Ir. Nat. J. 11: 120-132.

Bunce, R. G. H., 1982. A Field Key For Classifying British Woodland Vegetation. Part 1. Institute of Terrestrial Ecolo- gy, Cambridge, 103 pp.

Bunce, R. G. H. & Shaw, M. W., 1973. A standardised procedure for ecological survey. J. Environ. Manage. 1: 239-258.

Cooper, A., 1984a. Application of multivariate methods to a study of community composition and structure in an escarp- ment woodland in northeast Ireland. Vegetatio 55: 93-104.

Cooper, A., 1984b. Application of multivariate methods to the conservation management of hazel scrub in northeast Ire- land. Biol. Conserv. 30: 341-357.

Ellenberg, H., 1974. Indicator values for vascular plants in Cen- tral Europe. Scripta Geobotanica 9: 1-97.

Gauch, H. G., 1977. ORDIFLEX. Cornell University, New York, 185 pp.

Hill, M. O., 1979a. TWINSPAN. Cornell University, New York, 90 pp.

Hill, M. O., 1979b. DECORANA. Cornell University, New York, 52 pp.

Jermy, A. C. & Crabbe, J. A., 1978. The Island of Mull. London, British Museum (Natural History).

Kelly, D. L., 1975. Native woodland in western Ireland with es- pecial reference to the region of Killarney. Ph.D. thesis, University of Dublin, 337 pp.

Kelly, D. L. & Kirby, E. N., 1982. Irish native woodlands over limestone. J. Life Sci. R. Dubl. Soc. 3: 181-198.

Kelly, D. L. & Moore, J. J., 1974. A preliminary sketch of the Irish acidophilous oakwoods. Coll. Phytosoc. 3: 375-387.

K16tzli, E, 1970. Eichen-, Edellaub und Bruchwalder der Britischen Inseln. Schweiz. Z. Forstwes. 121: 329-366.

McConaghy, S. & McAleese, D. M., 1957. Studies of the basaltic soils of Northern Ireland 1. Cation-exchange properties. J. Soil Sci. 8: 127-134.

McVean, D. N. & Ratcliffe, D. A., 1962. Plant Communities of the Scottish Highlands. London, H.M.S.O.

Parker, R. E., 1979. Introductory Statistics for Biology. Arnold, London, 122 pp.

Peterken, G. E, 1974. A method for assessing woodland flora for conservation using indicator species. Biol. Conserv. 6: 239- 245.

Rackham, O., 1981. Ancient Woods. Arnold, London, 402 pp. Singer, S. B., 1980. DATAEDIT. Cornell University, New York,

42 pp.

74

Sokal, R. R. & Rohlf, F. J., 1981. Biometry. Freeman, San Fran- cisco, 859 pp.

Walkley, A. & Black, I. A., 1934. An examination of the Degt- jareff method for determining soil organic matter and a pro- posed modification of the chromic acid titration method. Soil. Sci. 37: 29-38.

Watson, E. V., 1968. British Mosses and Liverworts. Cambridge

University Press, Cambridge, 495 pp. Webb, D. A., 1977. An Irish Flora. Dundalgan Press, Dundalk,

277 pp.

Accepted 29.3.1985.