Introduction

The degree to which vegetation has been altered by human intervention is becoming more important in rating the conservation value of forest resources throughout the developed countries of the Northern Hemisphere (Franklin 1988; Seymour and Hunter 1999). Thus, based on the concepts of Jalas (1955) and Blume and Sukopp (1976), the hemeroby (ie, the reciprocal of naturalness) of forests in Austria was recently assessed in a Man and the Biosphere project (Grabherr et al 1998). Anthropogenic changes in tree layer species composition are regarded as a key attribute of hemeroby, ranking above understorey composition and diversity. In fact, changes in canopy dominance are likely to have impacts on many ecosystem properties, such as the diversity of the entire biotic community and soil quality. However plausible these impacts may be, scientific evidence for them is not easy to obtain because tree species impacts are longterm and often confounded by differences in primary ecological factors (Ulrich et al 1971; Binkley 1995).

If man-made changes in tree species composition alter understorey composition, the use of indicator plants (Ellenberg et al 1991) and character species (Braun-Blanquet 1964; Oberdorfer 1994) in the mapping of potential natural vegetation may be severely hindered. This is of particular relevance in the Alps, where vegetation is widely used in forest site classification (Schmider and Burnand 1988; Frey 1995; Ewald 1999).

At the northern fringe of the Alps, montane forests ranging from 600 to 1400 m were composed of European beech (Fagus sylvatica L.), Norway spruce (Picea abies [L.] Karst.), and silver fir (Abies alba Mill.) prior to human interference (Zukrigl 1973; Ewald 1997; Keller et al 1998). Palynological and historical records indicate that, in many parts of the northern Alps, tree species composition was altered in favor of Norway spruce in historical times (Von Bülow 1962; Kral 1979; Burga and Perret 1998). Meister (1969) identified selective browsing by deer that increased in density due to game preservation, forest pasturing, clear-cutting, artificial seeding, and planting as the causal agents behind this secular process. Such a change in canopy composition alters the living conditions of understorey plants, as mixed forests are replaced by pure conifer stands. Instead of nutrient-rich deciduous litter, spruce needles rich in polyphenols come to dominate the physical and chemical properties of the forest floor (Wittich 1961). Furthermore, coniferous canopies with their higher leaf area index and evergreen foliage filter larger amounts of atmogenic acids from the air (Van Breemen et al 1983; Kreutzer et al 1986). The conversion to Norway spruce is therefore likely to speed up soil acidification and induce the accumulation of poorly digested forest floor material, as Nihlgard (1971) showed in southern Swedish plantations. As a result, native understorey species with high demands on nutrients may decline in favor of acidophilous vegetation (Nihlgard 1970). The same effect was found in the foreland of the Swiss Alps (Teuscher 1985). Ewald (1997) hypothesized that spruce dominance favors character species of coniferous forests, which entails a danger of incorrect classification of potential natural vegetation. Recent studies in the Black Forest (Bürger 1991) and in northern Germany (Lücke and Schmidt 1997) indicate that artificial spruce stands have been subjected to a particularly rapid immigration of indicators of eutrophication due to elevated atmospheric nitrogen inputs.

Examining a representative sample of forest vegetation offers the chance to investigate which changes in species composition occurred in the montane belt of the Bavarian Alps. The present study aims to test hypotheses that tree stands with currently high proportions of Norway spruce could possess (1) fewer character species of temperate deciduous forests, (2) more character species of boreal/subalpine coniferous forests, (3) more plant indicators of acid conditions, and (4) more indicators of high nitrogen supply than mixed coniferous-deciduous and pure deciduous stands at comparable sites.

In order to remove the influence of site conditions that vary independent of the dominant tree species, the influence of environmental covariables was controlled in a partial correlation analysis (Legendre and Legendre 1998). This statistical method allows isolation of the ecological effect of canopy species across the wide range of environments present in a complex mountain landscape and prevents misinterpretations resulting from the nonrandom distribution of stand types across site types.

Material and methods

The vegetation sample was drawn from a monitoring network (Ewald and Reuther 1999) situated in the northern Calcareous Alps near Garmisch-Partenkirchen, Germany (11°4′–11°10′E, 47°24′–47°36′N). The montane forests studied occur between elevations of 630 and 1400 m, above which they are replaced by subalpine Norway spruce and dwarf pine (Pinus mugo Turra). Climate is cool montane (mean annual temperatures of 4–7°C, precipitation 1400–ca 2000 mm; Fliri 1975). Geological parent materials are diverse and give rise to a wide variety of soil types. Limestones/dolomites favor rendzina-type shallow soils rich in carbonates (Rendzic Leptosols according to FAO-UNESCO 1997), mudstones/schists develop into clayey and often waterlogged deep soils (Eutric Cambisols and Gleysols), and sandstones/siliceous limestones give rise to sandy acid soils (Eutric to Dystic Cambisols, rarely Podsols). Historically, pasturing and logging have been well documented since the 14th century. Early settlements can even be traced back to Roman times (Hauke 1988). Since the 19th century, preserving and winter feeding have resulted in high populations of red deer (Cervus elaphus), which are regarded as the main cause of a widespread decline of mixed forest in favor of Norway spruce (Figure 1). As silver fir has suffered most from this development, present-day stand types vary mostly from pure spruce to spruce-beech and occasional pure beech.

The selection of sampling points was based on an extensive forest inventory (see Ewald and Reuther [1999] for details) that allowed for preselection of stands older than 60 years and for stratification into stand types (pure spruce, mixed, pure beech), soil groups (6 classes), altitudinal zones (below and above 1000 m), slope aspect (sun-exposed, shaded), and petrography (5 zones read from a geological map). Sample points were drawn at random from 84 strata, thus providing a balanced sample, in which intercorrelations between stratification criteria are minimized.

At each sampling point, the variables listed in Table 1 were recorded. The first group describes diversity, composition, and structure of understorey vegetation types that served as dependent variables in statistical analysis. The classification of species into groups of deciduous forest (Querco-Fagetea) and coniferous forest (Vaccinio-Piceetea) was taken from the supraregional phytosociological study by Ewald (1997, Table 2). The number of species belonging to sociological groups was counted and cover values were summed per plot, once for all species and once for vascular plant species only. Understorey richness was calculated for all species and for vascular plants, bryophytes (mosses and liverworts), and tree regeneration (seedlings and saplings) separately. Total cover of vascular field layer and bryophyte layer were estimated visually in the field. Indicator values were taken from the list of Ellenberg et al (1991) and calculated as the simple mean of all species occurring in each 144-m² understorey sampling plot. Ellenberg's indicator value R expresses the ecological preference of plant species for soil pH on an ordinal scale from 1 to 9. Similarly, Ellenberg's N arranges plants with respect to their preference for varying levels of nitrogen availability. The arithmetic mean, mR, was calculated both with and without inclusion of bryophytes.

The explaining variable describing the proportion of spruce cover in the canopy (PropPice) was calculated as the percentage of Picea cover in the cover sum of all tree species recorded on a 24 × 24-m plot comprising the understorey sample at its center. Depth of mineral soil development and free carbonate were used as covariables since these soil properties—contrary to the dependent topsoil attributes—are not likely to be influenced by changes of canopy composition. They thus represent the component of water-retention capacity and trophic level that does not depend on the present state of the ecosystem. To account for the edaphic variability typical of mountain environments, soil profiles were inspected in 4 replicate augerings at 4 m from the plot center, and plot medians were used in statistical analysis. Slope exposition was transformed from angles to a favorability index (Beers et al 1966) by defining 157.5° as the most favorable thermical aspect. The total horizontal cover of the tree layer was included as a covariable to remove differences in stand management (thinning) that tend to reduce canopy density of Norway spruce stands in accessible locations.

Partial correlation analysis was carried out with SPSS for WINDOWS 9.0.1 software. PropPice was related to individual dependent variables while controlling for any linear effect of covariables. Partial correlation coefficients were tested for two-tailed significance.

Results

Besides giving an account of partial correlations and their significance (Table 3), this section shows the interplay between covariables and canopy composition in controlling vegetation attributes (Figures 2–4). The covariables with the highest simple correlation with the study variable of interest were selected for these scatterplots.

Of the 7 understorey cover variables (Table 3), only the bryophyte layer (CoveBryo) showed a significant positive partial response to the proportion of Norway spruce in the canopy (PropPice). Figure 2 shows that bryophyte cover increases both under spruce canopy and on flatter slopes. The significant partial relationship is due to the fact that high bryophyte cover on steep slopes was only encountered in stands dominated by spruce. Among the species-richness variables, the coniferous forest (Vaccinio-Piceetea) species with and without bryophytes and bryophyte richness respond positively to the partial influence of spruce canopy. Apart from being influenced by the canopy, bryophyte richness is most closely related to elevation (Figure 3).

The mean indicator value for soil reaction exhibits a negative partial relationship with PropPice, which becomes less pronounced when bryophytes are omitted. It can be read from Figure 4 that the differences in mR between deciduous and coniferous stands occur on soils with intermediate and great depths of free carbonate, whereas the occurrence of acid indicators on soils rich in carbonate was not related to canopy composition. There is no significant relationship between the indicator value for nitrogen and canopy composition. The negative correlation coefficient indicates that mN tends to be somewhat lower in stands dominated by spruce.

Conclusions and discussion

The significant partial effects of Norway spruce on vegetation suggest that hypotheses 2 and 3 hold up in the Bavarian Alps. Coniferous forest species and indicators of acidity occur in greater numbers in spruce stands than in mixed and deciduous stands at comparable sites. Lower pH in the uppermost humus layers may be ascribed to the higher release of organic acids from decomposing needle litter and, possibly, to higher atmogenic inputs of acids by canopy filtering (Ewald 2000). Contrary to expectations, the richness and abundance of deciduous forest species and the indicator values for nitrogen varied independently of canopy composition, so hypotheses 1 and 4 must be rejected.

The most pronounced effect on vegetation, which is not restricted to species richness but also has an effect on dominance structure, is on bryophytes. Bryophytes are clearly more abundant and more diverse in the presence of spruce. This corresponds to the findings of Lücke and Schmidt (1997) on artificial plantations in the Solling forest (northern Germany). Experiments by Longton and Greene (1979) showed that growth of the moss Pleurozium schreberi (Brid.) Mitt. was inhibited by the shading effect of broad-leaved litter. This suggests that the slow-growing bryophytes are simply relieved from inhibition by beech litter in coniferous stands. The vascular Vaccinio-Piceetea species are more numerous but not more dominant in spruce stands. Because common “raw humus species” such as Vaccinium myrtillus L. tend to have shallow root systems (Polomski and Kuhn 1998), they are more likely to respond to the observed superficial changes in soil chemistry than deciduous forest species. Thanks to the weak response of vascular species composition, floristically defined community types can be delimited and understorey plants can be used as ecological indicators regardless of past changes in tree species composition. However, the occurrence of Vaccinio-Piceetea species and bryophytes in particular can lead to misclassifications in phytosociological site mapping, which should always be complemented by physiographic methods of soil and relief interpretation (Ewald 1999).

As the comparison between the two modes of calculation indicates, the decrease in the reaction indicator value under Norway spruce is mostly due to acidophilous bryophytes. Because Ellenberg's indicator values are based on empirical induction from the ecological behavior of plants in the field (including plant sociology), the effect is conceptually related to the increase in Vaccinio-Piceetea species. Insignificantly lower values of mN in spruce stands, which correspond to larger C/N ratios in the uppermost soil (Ewald 2000), show that the findings from the Black Forest (Bürger 1991) and Solling (Lücke and Schmidt 1997) are not valid in the part of the Alps under consideration here. The absence of increased eutrophication in spruce stands may be due to relatively low inputs of nitrogen from agriculture and to relatively low levels of disturbance by forest management.

Some important conclusions can be drawn for the ecological evaluation of present forest composition in the study area that are likely to be valid in adjacent portions of the northern Calcareous Alps in Austria and Switzerland. As there is no measurable effect of spruce stands on understorey species richness in general and on deciduous forest species in particular, spruce stands do not cause overall impoverishment of the plant communities. Of course, this does not necessarily apply to other guilds and taxonomic groups such as epiphytes, fungi, insects, and mammals. Moreover, there are many good aesthetic, silvicultural, and economic reasons for arguing against single-species tree stands. Sociological and biogeographical species groupings such as “deciduous forest plants” (Querco-Fagetea) must be interpreted with care, as there appears to be no stringent functional link to deciduous forest ecosystems.