Pedunculate oak (Quercus robur L.) silviculture in natural stands of NW Spain: Environmental conditioners

Forest Ecology and Management 256 (2008) 702–711

Contents lists available at ScienceDirect

Forest Ecology and Management journal homepage: www.elsevier.com/locate/foreco

Pedunculate oak (Quercus robur L.) silviculture in natural stands of NW Spain: Environmental conditioners I.J. Dı´az-Maroto *, P. Vila-Lameiro Department of Agroforestry Engineering, University of Santiago de Compostela, Campus Universitario s/n, E-27002 Lugo, Spain

A R T I C L E I N F O

A B S T R A C T

Article history: Received 17 January 2008 Received in revised form 15 May 2008 Accepted 17 May 2008

Forest management is often carried out in different ways, without any appropriate environmental restrictions. Stands of pedunculate oak (Quercus robur L.) in Galicia (NW Spain) have been harvested by alternating high forest and, mainly, coppice forest. However, some totally inappropriate silvicultural treatments have been used, such as thinning of the best trees and inadequate pruning. The objective of the present study was to analyse how environmental characteristics affect the management of oak forests in Galicia. For this, a botanical inventory was carried out in 39 selected stands of Q. robur and a total of 42 parameters were measured, 4 of which were physiographical, 12 climatic, 19 edaphic and 7 silvicultural. In order to analyse the possible relationships among these variables, the silvicultural data were compared with the other data, by canonical correlation analysis. All parameters were correlated with the silvicultural regime, although the correlation was weak for the floristic data. It is therefore evident that the environmental conditions affect how forest stands should be managed, although this does not imply that more profitable use of the stands cannot be achieved than at present, and alternative silvicultural methods must be found to enable appropriate management and conservation of oak stands. ß 2008 Elsevier B.V. All rights reserved.

Keywords: Forest management High forest Coppice Quercus robur NW Spain

1. Introduction Forest management has a considerable influence on the stability and sustainability of forest ecosystems (Larsen, 1995; Johnson et al., 2002; Decocq et al., 2004). In the specific case of oak trees, the silvicultural practices applied form part of forestry mythology and should be considered with prudence (Bouchon and Trencia, 1990). The silviculture applied to Quercus robur L. are well developed and up to date in countries where the species is of great economic importance, such as France and Germany, where thousands of hectares of oak forest have been managed for centuries (Bary-Lenger and Nebout, 1993; Timbal and Aussenac, 1996; Harmer and Morgan, 2007). However, the situation in the study area is very different from that in the above-mentioned countries, as there is little knowledge about the type of silvicultural practices that should be applied to autochthonous broadleaf forests. Galician oaks present a wide range of ages and qualities, as a result of the different uses and states of conservation (Dı´az-Maroto et al., 2005). Coppice forest predominates and it requires continual management otherwise the stands will age and stagnate and may

* Corresponding author. Tel.: +34 9822 85900; fax: +34 9822 85926. E-mail address: [email protected] (I.J. Dı´az-Maroto). 0378-1127/$ – see front matter ß 2008 Elsevier B.V. All rights reserved. doi:10.1016/j.foreco.2008.05.023

eventually disappear (Dı´az-Maroto et al., 2006; Van Calster et al., 2007). Many of these oak forests have been intensively exploited (i.e. for the wood and firewood extractions for domestic and industrial uses or for the naval industry) (Manuel and Gil, 2001), and in many cases inappropriate silvicultural treatments have been applied (pollarding and felling of the best trees) (Dı´az-Maroto et al., 2005). The shipbuilding was prompted by the Spanish maritime expeditions, the commerce with Europe and the Indies, and the fishing activity. Due to transportation difficulties, the naval industry needed closer forests to fulfill their requirements. The wood proportion by ship was about 30–50% conifers, and about a 70–50% broadleaves, mainly oaks. The utilized trees had to have specific dimensions and they were of the best quality, and after their extraction, the rest was generally used for firewood or for vegetable coal, which contributed to the destruction of many forests (Aranda, 1990). Most recently, as a result of rural depopulation, technological developments and social demands, there has been a change during the past century from overexploitation of many of these forests to a total lack of silvicultural intervention (Izco et al., 1990; Roda´ et al., 1999; Reque and Bravo, 2007). Oak forests (carballeiras), or pure stands of Q. robur, occupy an area of 187,789 ha, in Galicia, i.e. approximately 14% of the total

I.J. Dı´az-Maroto, P. Vila-Lameiro / Forest Ecology and Management 256 (2008) 702–711

forest area (DGCONA, 2001). Within the region, Q. robur behaves as a robust, light-demanding species, which does not tolerate shade at early stages of development and the seedlings of which languish quickly under cover. The largest oak stands are found on steep slopes, where they have survived largely because felling would be complicated owing to the topography (Ruiz de la Torre, 1991). Management of these stands must now be carried out in accordance with silvicultural criteria and any possible environmental restrictions (Decocq et al., 2004; Vila-Lameiro and Dı´azMaroto, 2005). The starting hypothesis for the present study is that the environmental conditions in the study area do not preclude the application of alternative silvicultural practices in the management and conservation of oak forests, in order to obtain more profitable outputs rather than with traditional methods. 2. Materials and methods 2.1. Study area and sampling The study area comprises the Autonomous Community of Galicia (NW Spain) (Fig. 1). The climate is generally oceanic, and the main characteristic is therefore high precipitation (Carballeira et al., 1983). The rocks in the area are siliceous (granites, schists, and slates), under deep soils, particularly cambisols (Dı´az-Maroto and Vila-Lameiro, 2006). The vegetation composition in the study area was examined in the Mapa Forestal de Espan˜a (Ruiz de la Torre, 1991), obtaining the existing vegetation mosaics where Q. robur is present, and the sampling zones were selected from within these, with the help of

703

information provided by forestry administration personnel and data reported in previous studies (Dı´az-Maroto, 1997). Representative oak stands by each zone were chosen to replant the inventory plots. The minimum area of the stands considered ranged between 0.5 and 1 ha, which avoided problems associated with the edge effect. The resulting network of 39 rectangular plots (Table 1), of variable dimension (depending on the number of trees), contained at least 50 inventoriable trees (Ø  5 cm) (Hummel et al., 1959). 2.2. Variables measured and parameters calculated Once the plots were replanted, a floristic inventory was carried out and an abundance index was assigned to each (BraunBlanquet, 1979); abiotic (physiographic, climatic and edaphic) and biotic (dendrometric and silvicultural variables) data were then recorded. The physiographic and climatic data – adapted to each plot according to the method of Carballeira et al. (1983), and for the period 1960–2000 – and the results of the edaphic analyses – corresponding to the samples obtained from the soil profiles – enabled compilation of a total of 35 abiotic parameters for each plot (Table 2). The parameters that describe the physiography of each area (Dı´az-Maroto et al., 2005, 2006) are: altitude and mean slope, soil depth and distance from the sea. The climate was described by the following parameters (Retuerto and Carballeira, 1990, 1991): precipitation: annual, winter, spring, summer and autumn total values; temperature: mean annual, annual mean of absolute maximum and minimum temperatures, maximum value of the absolute maximum mean temperatures and minimum value of the absolute minimum mean temperatures (mTAB), and annual

Fig. 1. Distribution of the stands in the study area, Autonomous Community of Galicia, NW Spain.

I.J. Dı´az-Maroto, P. Vila-Lameiro / Forest Ecology and Management 256 (2008) 702–711

704

Table 1 Ecological and silvicultural features of each plot Plot

Altitude (m)

Slope (%)

Orientation

Parent rock

Texture

Type of soila

APb (mm)

MTb (8C)

NT (N0 ha

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39

650 240 1075 565 740 510 420 393 400 320 480 620 220 110 790 60 500 800 670 540 620 630 690 380 360 500 80 540 550 600 1300 600 520 300 1080 265 740 600 565

0 14 60 10 12 8 17 55 60 20 27 48 27 8 53 8 48 72 58 60 0 15 25 65 21 0 12 38 22 12 35 27 0 10 36 23 25 19 7

– North North Northwest Southwest Northeast North Northwest East Northeast South Southeast North Southwest North Northwest Northwest Northeast Northeast Northeast – Northwest North Northeast Southeast – Northeast Northwest Northwest North North Northwest – East East Southeast Southeast Northeast Northwest

Granite Schist Quartzite Granite Quartzite Granodiorite Granite Schist Schist Granite Granite Granite Granite Granodiorite Granite Gneiss Schist Schist/quartzite Schist/quartzite Schist Granite Granite Granite Granite Schist Granite Granite Schist Schist Schist Slates Granite Gneiss Slates/quartzite Granite Granodiorite Granite Schist Granite

Sandy Sandy Sandy-loam Sandy Sandy Sandy Loamy-sand Loamy-sand Loamy-sand Sandy-clay Sandy-clay Sandy Sandy Sandy Sandy Sandy Sandy-loam Sandy-loam Loam Loam Sandy Sandy-loam Sandy-loam Sandy Sandy-loam Sandy Sandy Loam Loam Loam Sandy-loam Sandy Sandy Loam Sandy Sandy-clay Sandy Sandy Sandy

Dystric cambisol Dystric cambisol Dystric cambisol Dystric cambisol Dystric cambisol Dystric cambisol Dystric cambisol Gleyic cambisol Humic umbrisol Dystric cambisol Humic umbrisol Humic umbrisol Dystric regosol Dystric regosol Humic regosol Dystric cambisol Dystric cambisol Dystric regosol Humic regosol Humic regosol Dystric cambisol Dystric cambisol Dystric cambisol Dystric cambisol Dystric cambisol Dystric cambisol Dystric cambisol Dystric cambisol Dystric cambisol Dystric cambisol Dystric cambisol Dystric cambisol Dystric cambisol Humic umbrisol Humic regosol Dystric cambisol Dystric cambisol Dystric cambisol Dystric cambisol

867 1485 1392 841 1947 1178 1169 1684 1684 1715 1574 1645 772 1345 1276 1715 1584 1116 1116 1116 1194 1194 1194 1424 1574 1089 1485 1130 1130 1130 1176 1492 1574 1708 1125 1345 1947 1492 841

10.6 13.5 9.5 11.0 12.0 11.3 11.5 11.7 11.7 14.1 11.9 9.6 12.8 12.4 12.9 14.1 11.7 9.9 9.9 9.9 10.4 10.4 10.4 11.1 11.9 11.3 14.6 11.6 11.6 11.6 7.3 11.2 11.9 10.8 8.6 12.4 12.0 11.2 11.0

500 867 1563 1033 1167 767 567 900 667 1467 667 933 767 767 433 533 667 749 403 330 1853 2059 1956 1167 333 486 333 1583 1362 1453 453 967 1900 767 2000 833 900 1067 1167

1

)

BA (m2 ha

1

)

12.07 66.40 40.21 128.22 55.03 29.51 12.44 100.35 18.69 39.37 42.28 74.32 63.84 93.79 63.55 74.40 47.48 24.33 21.50 16.96 19.69 21.88 20.78 106.98 11.14 13.26 66.20 25.45 25.18 25.33 65.53 171.81 48.63 25.54 26.52 112.60 70.14 35.20 101.65

AP: total annual precipitation; MT: mean annual temperature; NT: number of trees per hectare; BA: basal area per hectare. a According to FAO (2006). b Climatic data period: 1960–2000.

mean of the maximum and minimum temperatures. To evaluate the soil properties and fertility, the following edaphic parameters were considered (Rubio et al., 1997; Roda´ et al., 1999; Dı´az-Maroto and Vila-Lameiro, 2006): pH in H2O, organic matter, nitrogen, carbon–nitrogen ratio, available phosphorus, exchangeable potassium, calcium and magnesium, sand, loam and clay. For all edaphic parameters, the total value was calculated from the weighted mean for the whole profile, by the method of Russell and Moore (1968). As regards the surface values, the data from the edaphic epipedon (upper 20 cm of the profile) were used, except when there was more than one horizon in this profile, in which case a weighted mean was calculated. For sand, loam and clay, only the total values were calculated. The structure of the Q. robur stands and the silvicultural treatments carried out were evaluated by use of the following dendrometric/silvicultural parameters (Table 2) (Rubio et al., 1997; Dı´az-Maroto et al., 2005, 2006): tree density, basal area (BA), coefficient of variation of the diameter distribution, coefficient of variation of the height distribution, Assmann’s dominant height (ADH) (Assmann, 1970), Hart’s index (HAI) (which estimates the distribution of trees in relation to their dominant height), and Czarnowski’s index (CZI) (number of trees in a squared plot of side equal to the arithmetic mean height) (Rondeux, 1993). The basic condition that the dendrometric/silvicultural variables must meet for use as possible discriminant parameters is that they are

independent of age and site quality, thereby allowing comparison of stands of different ages and qualities (Swanson and Franklin, 1992). In this sense, the coefficients of variation are nondimensional, and the Hart and Czarnowski indices are relatively independent of age and site quality. The same does not apply to the other three stand parameters, however the number of trees, the BA and the dominant height are fundamental for defining the structure of the stands studied (Rondeux, 1993; Rubio et al., 1997). 2.3. Statistical analysis Firstly, the most relevant descriptive statistics were calculated (Johnson, 2005) for each of the parameters studied, with the aim of analyzing the distribution and dispersion of the values (SAS Institute Inc., 2004). Then, to evaluate the relationship between the silvicultural parameters and the abiotic parameters divided among three groups of data – physiographic, climatic and edaphic – and also the floristic data, a canonical correlation analysis (CCA) was applied starting with a discriminant analysis for classifying cases. This tool enables definition of a linear combination of independent environmental variables, denominated as discriminant functions, which allow classification of each of the stands in some of the subpopulations or groups established by the values of the dependent variable of the grouping (ter Braak, 1994; SAS Institute Inc., 2004). One of the limitations of CCA is the number of

I.J. Dı´az-Maroto, P. Vila-Lameiro / Forest Ecology and Management 256 (2008) 702–711 Table 2 Abiotic and biotic parameters considered in each plot

Table 3 Descriptive statistics for the parameters under study

Number

Parameter (units)

Code

1 2 3 4 5 6 7 8 9 10 11 12 13

Mean altitude (m) Mean slope (%) Soil depth to the parent rock (cm) Closest distance from the sea (km) Total annual precipitation (mm) Winter precipitation (mm) Spring precipitation (mm) Summer precipitation (mm) Autumn precipitation (mm) Mean annual temperature (8C) Annual mean of absolute maximum temperatures (8C) Annual mean of absolute minimum temperatures (8C) Maximum of the absolute maximum mean temperatures (8C) Minimum of the absolute minimum mean temperatures (8C) Annual mean of the maximum temperatures (8C) Annual mean of the minimum temperatures (8C) Total pH in H2O Surface pH in H2O Total organic matter (%) Surface organic matter (%) Total nitrogen (%) Surface nitrogen (%) Total C/N ratio Surface C/N ratio Total available phosphorus (ppm) Surface available phosphorus (ppm) Total exchangeable potassium (ppm) Surface exchangeable potassium (ppm) Total exchangeable calcium (ppm) Surface exchangeable calcium (ppm) Total exchangeable magnesium (ppm) Surface exchangeable magnesium (ppm) Total sand (%) Total silt (%) Total clay (%) Tree density: number of trees per hectare (N0 ha 1) Basal area per hectare (m2 ha 1) Coefficient of variation of the diameter distribution (%) Coefficient of variation of the height distribution (%) Assmann’s dominant height (m) Hart’s index (%) Czarnowski’s index

ALT SLP DPTH DS AP WP SpP SuP AuP MT AMT AmT MTAB

14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42

705

mTAB MTM MTm PH SPH OM SOM N SN C/N SC/N P SP K SK Ca SCa Mg SMg Sand Silt Clay NT BA CVD CVH ADH HAI CZI

independent variables used in the analysis, as it is recommended that the number should not exceed 30 (Gittins, 1985; Jongman et al., 1995). To resolve this problem, the values of the grouping variable, in this case, the silvicultural data, should be discretely defined; for this, a principal component analysis (PCA) was first applied to summarise all of the silvicultural information in a single vector, which was then reclassified for each plot on a scale between 1 and 5, to ensure uniformly sized and not excessively numerous blocks of variables (SAS Institute Inc., 2004; Johnson, 2005). The most serious problem with the independent variables arose with the data floristic, and to overcome this they were also first summarised by PCA on the basis of the cover scale values of Braun-Blanquet (1979), and grouped into six pseudo-species, which represent ranks of relative abundance of species, identified with the same name followed by a number: zero (absence), one (up to 10% cover), two (10–25%), three (25–50%), four (50–75%) and five (>75%) (Dı´az-Maroto and Vila-Lameiro, 2007). Finally, to determine the statistical relevance of the vectors resulting from the discriminant analysis, the most representative of each group of variables (physiographic, climatic, edaphic and floristic) were selected and the bivariate correlations between these and silvicultural variables were used to identify the

Parameter (units)

Range

Mean  S.D.

CV (%)

Skewness

Kurtosis

ALT (m) SLP (%) DPTH (cm) DS (km) AP (mm) WP (mm) SpP (mm) SuP (mm) AuP (mm) MT (8C) AMT (8C) AmT (8C) MTAB (8C) mTAB (8C) MTM (8C) MTm (8C) PH SPH OM (%) SOM (%) N (%) SN (%) C/N SC/N P (ppm) SP (ppm) K (ppm) SK (ppm) Ca (ppm) SCa (ppm) Mg (ppm) SMg (ppm) Sand (%) Silt (%) Clay (%) NT (N0 ha 1) BA (m2 ha 1) CVD (%) CVH (%) ADH (m) HAI (%) CZI

60–1300 0–72 46–150 1–135 772–1947 226–645 151–375 61–283 281–644 7.3–14.6 20.0–28.8 4.3–6.2 27.3–38.5 11.5–1.7 11.7–20.7 2.1–10.5 3.92–6.15 3.82–6.53 1.04–23.31 1.19–34.21 0.042–0.793 0.050–1.019 6.9–29.6 10.4–30.1 0.4–117.2 0.4–119.5 9–231 19–252 3–1297 4–1704 0–85 0–143 13.83–88.67 7.44–84.89 1.27–26.82 333  2059 11.14–171.81 17.49–55.34 15.22–31.44 9.28–28.87 10.14–33.42 6.92–71.00

539  260 27  21 94  26 42  28 1372  311 436  120 269  53 164  51 481  108 11.5  – 24.0  – 0.8  – 32.6  – 4.2  – 16.7  – 6.7  – 4.85  0.46 4.71  0.51 8.64  5.19 12.85  7.76 0.307  0.178 0.442  0.232 14.6  4.5 16.8  4.3 21.8  29.1 19.8  29.2 73  40 103  50 120  216 170  285 29  21 49  28 65.78  20.50 22.53  20.75 11.22  5.09 935  474 57.00  41.27 34.74  10.94 21.86  4.90 17.96  4.44 20.75  6.18 22.32  15.65

48.2 77.7 27.6 66.6 22.7 27.5 19.7 31.1 22.5 – – – – – – – 9.3 10.7 60.0 60.4 58.0 52.5 31.0 25.3 133.6 147.5 54.8 48.5 180.0 167.6 72.4 77.5 31.2 92.1 45.4 50.7 72.4 31.5 22.4 24.7 29.8 70.1

0.64 0.59 0.47 1.25 0.14 0.13 0.02 0.47 0.15 – – – – – – – 0.60 1.27 0.70 0.82 1.08 0.35 1.02 1.60 2.01 2.52 1.79 0.65 4.52 4.39 1.05 0.89 0.70 0.47 0.41 0.76 1.15 0.27 0.32 0.59 0.29 1.89

1.32 0.84 0.43 2.21 0.54 1.18 0.35 0.17 1.18 – – – – – – – 0.97 3.55 0.23 0.46 1.19 0.48 2.32 2.55 3.36 5.62 5.38 0.74 24.04 22.81 0.80 0.06 0.23 0.43 0.03 0.43 1.02 1.06 1.14 0.93 0.69 3.81

S.D.: standard deviation; CV: coefficient of variation.

statistically significant relationships, with the maximum admissible error corresponding to a significance level a = 0.05 (Johnson, 2005). 3. Results The descriptive statistics for the 42 parameters considered are shown in Table 3. The Q. robur stands are situated in areas with a range of slopes of between 0% and 72% and of altitudes of between 60 and 1300 m, the latter which is reached in the mountainous zone in the east of Galicia (Os Ancares and O Courel); the distance from sea ranges from 1 to 135 km. The climate is characterized by high precipitation, with values higher than 700 mm in all stands, and close to 2000 mm in some. The summer precipitation is usually higher than 120 mm; those oak stands in which such high values are not reached are compensated by horizontal precipitation in the form of mist. The mean annual temperature is 11.5 8C, the maximum value of the absolute maximum mean temperatures is 32.6 8C, with a maximum temperature of 38.5 8C, whereas the minimum value of the mTAB is 4.2 8C, with a minimum temperature of 11.5 8C (Table 3). In the PCA of all discretely defined stand parameters (tree density, BA, coefficient of variation of the diameter and height

I.J. Dı´az-Maroto, P. Vila-Lameiro / Forest Ecology and Management 256 (2008) 702–711

706 Table 4 Results of PCA of the stand parameters KMO test

Stand parameters

0.840

Variable

Variability explained (%)

Accumulated variability (%)

1

98.4

98.4

2 3

1.2 0.4

99.6 100.0

Explanatory parameters

NT, BA, HAI, and CZI CVD and CVH ADH, HAI, and CZI

Table 6 Results of the test of significance for the variability intra- and intergroups of parameters Type of parameter

Parameter

lW

Sig F

Physiographic

ALT SLP DS

0.792 0.949 0.790

0.164 0.831 0.159

Floristic

Quercus robur Rubus spp. Pteridium aquilinum Hedera helix Lonicera periclymenum Erica arborea Teucrium scorodonia Vaccinium myrtillus Pyrus cordata Castanea sativa Ilex aquifolium Holcus mollis Agrostis capillaris Ruscus aculeatus Melampyrum pratense Asphodelus albus Betula alba Stellaria holostea Anthoxanthum odoratum SAI = 16–25 SAI = 11–15 SAI = 6–10 SAI = 5 SAI = 4 SAI = 3 SAI = 2 SAI = 1

0.705 0.765 0.753 0.902 0.543 0.678 0.722 0.839 0.729 0.911 0.876 0.887 0.884 0.979 0.672 0.553 0.840 0.497 0.795 0.750 0.757 0.856 0.860 0.771 0.662 0.800 0.947

0.038 0.101 0.084 0.557 0.001 0.024 0.052 0.280 0.058 0.610 0.430 0.481 0.469 0.961 0.021 0.002 0.281 0.000 0.156 0.091 0.089 0.343 0.359 0.110 0.018 0.166 0.811

Climatic

AP SuP MT AMT AmT MTAB MTM MTm mTAB

0.912 0.909 0.718 0.863 0.490 0.861 0.908 0.660 0.298

0.631 0.613 0.054 0.392 0.001 0.381 0.609 0.020 0.000

Edaphic

PH SPH MO SMO N SN C/N SC/N P SP K SK Ca SCa Mg SMg

0.545 0.661 0.784 0.806 0.781 0.807 0.988 0.962 0.775 0.792 0.883 0.872 0.903 0.884 0.841 0.863

0.002 0.021 0.147 0.196 0.141 0.198 0.986 0.898 0.129 0.164 0.480 0.428 0.581 0.484 0.304 0.391

KMO: Kaiser–Meyer–Olkin coefficient.

distribution, ADH, HAI and CZI, an optimal result was obtained, with a value of the Kaiser–Meyer–Olkin (KMO) coefficient >0.8, and a single vector that explained more than 98% of the variability in the silvicultural data (Table 4). As regards the discriminant analysis, the worst result from all of the groups of data analysed corresponded to the physiographic parameters (altitude, slope, soil depth and distance from the sea). Analysis of the Wilks l coefficient (lW), indicates that the sum of intragroup squares is very high, with lW close to 1, and therefore almost equal to the total sum of squares (Table 5). The parameters altitude and distance from the sea underwent an important change in significance, with an approximate error of 15% according to the value of the Sig F coefficient, compared with a value of more than 80% for slope and depth of soil (Table 6). The CCA explained 87% of the variability corresponding to the physiographic parameters, requiring two vectors, for which the canonical correlations with the silvicultural variables ranged between 0.45 and 0.50 (Table 5). In a bivariate correlation analysis, the first of these presented a statistically significant relationship with the silvicultural variable, tree density (NT). However, the second was correlated with silvicultural variables associated with tree spacing and stand height, such as ADH, HAI and CZI, as well as with the BA and the number of trees (Table 8). As regards the climatic data, the discriminant analysis differentiated two groups of variables. On one hand the pluviometric variables, with a very high intragroup variability (lW > 0.9) and on the other hand, thermal variables, for which the intergroup variability was generally higher than the intragroup variability, as the values of lW were lower and for some variables, the Sig F coefficient was close to zero (Table 6). On the basis of these results, in the CCA, only the first vector created was considered, as it explained 85% of the total variability and the value of lW was less than 0.03 (Table 5). In addition, this vector showed the most significant correlations with the silvicultural variables, specifically Table 5 Results of discriminant analysis (CCA) of stand and other parameters Type of parameter

Eigenvalue Variable Variability lW explained (%)

Canonical DF correlation

Sig F

Physiographic

0.323 0.281

1 2

46.5 40.5

0.541 0.494 0.715 0.468

16 0.432 9 0.448

Climatic

9.798 1.241

1 2

85.0 10.8

0.027 0.953 0.292 0.744

32 0.000 21 0.089

Edaphic

3.865 2.164 1.073

1 2 3

50.5 28.3 14.0

0.020 0.891 0.098 0.827 0.310 0.719

72 0.346 51 0.697 32 0.883

Floristic

211.394 57.011 11.560

1 2 3

74.7 20.1 4.1

0.000 0.998 0.000 0.991 0.020 0.959

104 0.000 75 0.000 48 0.053

lW: Wilks coefficient; DF: degrees of freedom; Sig F: significance of function.

lW: Wilks coefficient; Sig F: significance of function; SAI: sum of the abundance indices (Braun-Blanquet, 1979).

with CZI and with parameters related to tree spacing, such as ADH and HAI (Table 8). The intergroup variability of the edaphic parameters was highly significant, especially that for pH, and to a lesser extent, organic matter, nitrogen and phosphorus. In contrast, the highest intragroup sum of squares (lW  1) corresponded to the C/N ratio and, to a lesser extent, to the concentrations of potassium, calcium and magnesium (Table 6). The CCA of the silvicultural variables defined by the PCA were very significant, with values of between 0.70 and 0.90, and it was possible to explain 93% of the intergroup

I.J. Dı´az-Maroto, P. Vila-Lameiro / Forest Ecology and Management 256 (2008) 702–711

707

Table 7 Results of PCA of the floristic parameters

SAI = 1

SAI = 2

SAI = 3

SAI = 4

SAI = 5

SAI = 6–10

KMO test

Variable

Variability explained (%)

Accumulated variability (%)

Explanatory parameters

0.751

1

42.7

42.7

2

26.2

68.9

3

19.7

88.6

Circaea lutetiana Cirsium palustre Danthonia decumbens Juncus effusus Juncus sylvaticus Prunus spinosa Solanum dulcamara Arum italicum Hyacinthoides non-scripta Leontodon taraxicoides Plantago lanceolata Corydalis claviculata Lathyrus angulatus Valeriana pyrenaica

1

36.1

36.1

2

22.0

58.1

3

15.2

73.3

4

14.5

87.8

5

11.6

99.4

1

53.8

53.8

2

31.9

85.7

1

52.5

52.5

2

35.7

88.2

1

55.0

55.0

2

37.2

92.2

1

33.9

33.9

2

23.5

57.4

3

9.9

67.3

4

9.1

76.4

0.811

0.770

0.409

0.512

0.741

Athyrium filix-max Davallia canariensis Icopus europaeus Ligustrum vulgare Lamium maculatum Mercurialis perennis Polygonatum verticilatum Poa pratensis Veronica montana Rosa canina Vicia spp. Adenocarpus complicatus Arenaria montana Athyrium filix-femina Cynodon dactylon Silene dioca Valeriana montana Erica umbellata Oenanthe crocata Sedum anglicum Geranium robertianum Melica uniflora Acer pseudoplatanus Aquilegia vulgaris Ranunculus repens Dactylis glomerata Luzula campestris Veronica officinalis Tammus communis Arum maculatum Prunus padus Lotus uliginosus Sorbus aucuparia Omphalodes lusitanica Euphorbia dulcis Luzula sylvatica Oxalis acetosella Euphorbia amygdaloides Saxifraga spathularis Rumex acetosa Fagus sylvatica Arhenatherum elatius Dryopteris dilatata Primula vulgaris Oxalis acetosella Genista florida subsp. polygaliiphylla Calluna vulgaris Halimium alyssoides Potentilla erecta Erica cinerea Lotus corniculatus

I.J. Dı´az-Maroto, P. Vila-Lameiro / Forest Ecology and Management 256 (2008) 702–711

708 Table 7 (Continued )

SAI = 11–15

SAI = 16–25

KMO test

Variable

Variability explained (%)

Accumulated variability (%)

Explanatory parameters

0.442

1

33.4

33.4

2

24.6

58.0

3

12.7

70.7

Anemone nemorosa Festuca rubra Physospermum cornubiense Simenthis planifolia Agrostis delicatula Crataegus monogyna Viola riviniana Quercus pyrenaica Hypericum pulchrum Corylus avellana Agrostis curtisii Anemone trifolia Polygonatum odoratum

1

40.7

40.7

2

30.2

70.9

0.540

Pseudarrhenatherum longifolium Cytisus scoparius Daboecia cantabrica Lithodora postrata Blechnum spicant Polypodium vulgare Cytisus striatus

SAI: sum of the abundance indices (Braun-Blanquet, 1979).

edaphic variability with three vectors (Table 5). In the bivariate correlation analysis, all of these were significantly related to the same silvicultural variable, the BA, and were inversely proportional in vectors 2 and 3 (Table 8). The total number of species inventories was approximately 200. Owing to the limitations of CCA as regards the number of variables that it is able to manage, the presence of the species with minor abundance index (SAI) following the Braun-Blanquet (1979) methodology had to be grouped. As a result of this new distribution, eight groups were defined. The first group, SAI = 1, represent the count of all species which abundance amount in the context of the 39 plots result equal to 1 (zero in 38 plots and 1 in one plot). With the same argument another seven groups were defined, joining abundance indices equal to 2, 3, 4, 5, 6–10, 11–15 and 16–25. When the SAI result mayor to 25 the species were not grouped. As these species were the most abundant, were analysed individually. The five most abundant species were: Rubus spp., Pteridium aquilinum (L.) Kuhn, Hedera helix L., Lonicera periclymenum L. and Teucrium scorodonia L. These species, along with Erica arborea L., were the only ones present in more than 30 inventories. A PCA was carried out of each of the resulting groups (Table 7), and produced new variables or vectors with which to try to explain the variability caused by the different species. The KMO coefficient for the validity of the analysis was high in all cases, usually greater Table 8 Correlations between the discriminant analysis vectors and silvicultural parameters Eigenvalues (Table 5)

BA

NT

CVD

ADH

HAI

CZI

Physiographic 1 Physiographic 2

n.s. 0.345

0.427 0.437

n.s. n.s.

n.s. 0.384

n.s. 0.480

n.s. 0.478

Edaphic 1 Edaphic 2 Edaphic 3

0.464 0.405 0.418

n.s. n.s. n.s.

n.s. n.s. n.s.

n.s. n.s. n.s.

n.s. n.s. n.s.

n.s. n.s. n.s.

Climatic 1 Climatic 2

0.340 n.s.

n.s. n.s.

n.s. n.s.

0.423 n.s.

0.370 n.s.

0.445 n.s.

Floristic 1 Floristic 2 Floristic 3

0.489 n.s. n.s.

n.s. n.s. n.s.

n.s. 0.386 n.s.

n.s. n.s. n.s.

n.s. n.s. n.s.

n.s. n.s. n.s.

n.s.: correlation not statistically significant.

than 0.5. The information regarding the vegetation was summarised by the 19 most abundant species and the vectors derived from the PCA of the 8 groups characterized by the value of SAI (Tables 6 and 7), so that at least 70% of the initial variability was explained. The discriminant analysis of the vegetation provided good results, as 95% of the variability in the Q. robur stands was explained by only two vectors. Both vectors showed a canonical correlation of more than 0.9, i.e. a highly significant value, and a lW value of zero (Table 5). The species included in the first vector were: Castanea sativa Mill., Anthoxanthum odoratum L., Vaccinium myrtillus L., Agrostis capillaris L., Ruscus aculeatus L., Stellaria holostea L., Ilex aquifolium L. and, to a lesser extent, H. helix, Melampyrum pratense L., Holcus mollis L. and Pyrus cordata Desv. As regards the second vector, L. periclymenum, P. aquilinum, T. scorodonia, Rubus spp. and Betula alba L. had a noteworthy influence. The five most abundant species in addition to Q. robur (Rubus spp., P. aquilinum, H. helix, L. periclymenum and T. scorodonia) therefore showed a certain, although not very high, degree of discriminatory capacity. Despite the results obtained with these two vectors in the previous statistical analyses, both show few significant correlations with the silvicultural variables, with the only noteworthy relationships those between the first vector and the BA, and between the second vector and the coefficient of variation for the diameter distribution (Table 8). 4. Discussion and conclusions Environmental factors are undoubtedly important in terms of the growth and production of trees and forests (Larsen, 1995; Berges et al., 2006). However, the question remains as to how forest stands in general and Q. robur stands in particular respond to different silvicultural treatments within their ecological limits (Rubio et al., 1997; Guilley et al., 2004). Within the study area, Q. robur forests are climax formations that cover a large part of the region, from close to sea level to 1300 m (Dı´az-Maroto et al., 2005). Several silvicultural aspects influence the interpretation of their natural ecological limits. Firstly, pedunculate oak stands have been intensively managed to date, although with different objectives and criteria than those applied in other European countries (Valkonen et al., 2002; Bary-

I.J. Dı´az-Maroto, P. Vila-Lameiro / Forest Ecology and Management 256 (2008) 702–711

Lenger and Nebout, 1993; Decocq et al., 2004); secondly, the longterm dynamics of slow growing species depend on the extent and frequency of the silvicultural interventions (Aude and Lawesson, 1998; Aussenac, 2000). 4.1. Descriptive statistics for the ecological parameters The mean value of the slope (SLP) indicates that at present many oak forests are located in areas of complex topography, with steep slopes, and where they have remained because felling in such areas would be very difficult to carry out (Izco et al., 1990; Ruiz de la Torre, 1991). Although the values of the parameter DS (closest distance from the sea) in the stands under study are consistent with the potential distribution of vegetation in the northwest of the Iberian Peninsula (Izco et al., 1990; Rivas-Martı´nez et al., 2001; del Rio and Penas, 2006), some pedunculate oak stands have become established far from their potential area. In such cases, they have formed hybrids with other species of the genus Quercus, such as Quercus petraea Liebl. and Q. pyrenaica Willd. (Dı´az-Maroto and Vila-Lameiro, 2007). The general prevalence of a humid oceanic climate (Carballeira et al., 1983) explains the homogeneity of the climatic parameters, except the annual mean of absolute minimum temperatures (AmT) (Table 3). The variability in this parameter is due to the high thermal range, accentuated by the variation in altitude among stands (Retuerto and Carballeira, 1991). The high values of both total annual and summer precipitation are indicative of the moisture requirements of Q. robur (Timbal and Aussenac, 1996; Guchu et al., 2006). The existence of oak forests in zones where summer droughts occur, mainly in the province of Ourense (Fig. 1), where there is a certain Mediterranean influence in the climate (Carballeira et al., 1983), is possibly due to the moisture in the form of fog and mist. 4.2. Relation between silvicultural and other parameters As previously mentioned, to study the possible relationships between silvicultural and other parameters, a CCA was carried out, in an attempt to obtain a linear combination of independent environmental variables, which would enable classification of the stands into some of the groups established by the values of the dependent variable (silvicultural information) (Gittins, 1985; ter Braak, 1994; Jongman et al., 1995). For this, the silvicultural data were discretely defined and the application of PCA enabled them to be summarised in a single vector comprised by the stand parameters NT (tree density), BA, HAI and CZI (Table 4). 4.2.1. Relationship with abiotic parameters (physiographic, climatic and edaphic) The results of the discriminant analysis of the physiographic parameters demonstrate the non-existence of significant differences among the mean values, according to the model of grouping obtained for the silvicultural data. This also occurred with the CCA. However, the patterns observed for the parameters altitude and distance from the sea were not clear, and therefore given the importance that both of these have on the oceanic and continental influence in the studied stands (Retuerto and Carballeira, 1990; Aussenac, 2000), their effect in the discriminant analysis of the climatic parameters must be ratified. Given that the first vector resulting from the discriminant analysis of the physiographic parameters was significantly and inversely proportional to the silvicultural variable NT (Table 8), it was possible to analyse its effect on the continentality of the stands (Johnson et al., 2002; Decocq et al., 2004). The parameter tree density is related to tree spacing and thus to diameter and height.

709

Therefore, the lower the values of altitude, slope and distance from the sea – parameters that define the first physiographic vector (Table 6) – the greater the density and lower the spacing. This is also true for the second vector, which was correlated with the silvicultural variables ADH, HAI, CZI, BA and NT (Table 8), also linked to tree spacing and stand height. These results show that the best sites for Q. robur in northwest Spain correspond to locations of thalweg or intermediate slope, where the existence of physiographic characteristics that are more optimal for a species combine with other climatic characteristics that are also more favourable and that prevent natural substitution of these by other species such as Q. petraea or Fagus sylvatica L. (Kelly, 2002; Dı´az-Maroto et al., 2005). Taking into account the results of the discriminant analysis of the climatic parameters, with differentiation of a group of pluviometric variables (AP and SuP) and another of climatic variables (MT, AMT, AmT, MTAB, MTM, MTm and mTAB) (Table 6), only the first vector was considered in the CCA (Table 5), in which the greatest weighting corresponded to those parameters related to minimum temperatures, particularly the minimum of the mTAB and the annual mean of AmT. Both parameters exert an influence at the start and duration of the annual growth period of Q. robur (Timbal and Aussenac, 1996; Valkonen et al., 2002). Given that the vector is significantly correlated with several silvicultural variables (CZI, ADH, HAI and BA) (Table 8), related to tree spacing and stand height (Rondeux, 1993), we can verify that for lower values of mTAB and AmT, the slenderness coefficient (tree height/diameter) values were lower, as dominant tree height (ADH) was lower. On the basis of the above, we can confirm that in relation to climatic characteristics, the most suitable sites for pedunculate oak within the study area are those with the highest minimum temperatures, and the existence of a wide temperature range is not suitable; both of these factors are associated with the growth period of the species (Bary-Lenger and Nebout, 1993; Dı´az-Maroto et al., 2005). If in addition to these temperature conditions, both the annual and summer levels of precipitation are higher, growth of the stand in general and of the dominant trees in particular, will be optimal, with presence of a high degree of competition for light that converts into high values of slenderness coefficient, and therefore high mean and dominant heights (Bouchon and Trencia, 1990; Aude and Lawesson, 1998; Aussenac, 2000). The canonical correlations between the vectors resulting from the combination of the edaphic and silvicultural variables defined by the PCA were poorly significant, with pH, nitrogen, C/N ratio and organic matter being the parameters with importance, although the results obtained are not totally conclusive (Table 5). The deep soils on which the oak stands are established have developed on mainly siliceous and very similar substrates (Table 1), which probably gives rise to the significant relationships observed (Aussenac, 2000; Johnson et al., 2002). In the first of the vectors, the representative variables were pH and concentration of phosphorus, and the vector was significantly correlated with BA. The other two vectors, identified by the concentration of magnesium and C/N ratio (second) and by the organic matter and nitrogen contents (third) were also correlated with the BA, although the relationship was inversely proportional (Table 8). Therefore, in addition to greater stand growth – higher BA – soils with lower pH values, good mineralization of the vegetable remains and adequate incorporation of the organic matter could be required (Berges et al., 2006; Lo¨f et al., 2006); however, these conditions, except the low pH, are not found in Galician soils (Dı´azMaroto and Vila-Lameiro, 2006). The edaphic characteristics are of less importance in the silvicultural variables related to slenderness, tree spacing and stand height, as these factors are more limited by physiography and climate.

710

I.J. Dı´az-Maroto, P. Vila-Lameiro / Forest Ecology and Management 256 (2008) 702–711

4.2.2. Relationship with floristic data The CCA of the floristic data was very complicated because of the large number of species involved. Once the inventoried species were grouped, PCA was applied to each of the 8 groups considered, characterized by the value of SAI (Table 7). This procedure enabled summarising of the information regarding the vegetation, and on one hand into the 19 most abundant species (SAI > 25) and on the other hand, into the necessary vectors obtained from the PCA (Dı´az-Maroto and Vila-Lameiro, 2007). The discriminant analysis (CCA) of the floristic variables provided two new vectors that explained almost 95% of the existing variability (Table 5). However, both vectors were poorly correlated with the silvicultural parameters considered (Table 8), which may indicate that the growth and production of the pedunculate oak stands in the study area are not affect by accompanying vegetation, either at tree level or at lower levels (shrub, subshrub and herbaceous status) (Decocq et al., 2004; Tarrega et al., 2007; Van Calster et al., 2007). In fact, the only noteworthy correlation was that between the first vector and the BA (Table 8). In this vector, the following species were the most important: C. sativa, V. myrtillus, R. aculeatus, I. aquifolium, H. mollis and P. cordata, belonging all to the vegetation series associated with Q. robur and define different regression states of the potential vegetation (Rivas-Martı´nez et al., 2001), but do not determine stand growth. 4.3. Silvicultural aspects of pedunculate oak forests and management alternatives The silvicultural status of these stands is characterized by the following aspects (Dı´az-Maroto et al., 2005, 2006; del Rio and Penas, 2006; Diaci et al., 2008): (1) slow growth of the stands, scarce fruiting, accentuated by the mast fruiting of the species and conditions that are rather unfavourable for development of seedlings (Zaczek, 2002; Lo¨f et al., 2006); (2) aging of the strains that causes risk of vegetative decaying, in addition some stands present (Valkonen et al., 2002; Zaczek, 2002); (3) canopy closed with trees of small diameter, some showing top die back, and abundant scrub, which increases the risk of fire. Consequence of previous analysis, floristic data present a little influence in oak development, however, the management of site conditions resulted more important (Guilley et al., 2004; Diaci et al., 2008). Therefore, alternative management tools are required for Q. robur stands in the study area to enable more profitable production than obtained with traditional production methods; among such methods the following are possible: conversion of stands to high forest, maintenance of coppice forests in stands of a certain quality and homogeneity, which are still used for firewood production, silvopastoral improvement in areas where extensive grazing is important, and restoration of highly degraded stands by reforestation with other species of the genus Quercus, such as Q. petraea and Q. pyrenaica, or with other broadleaf species (Park, 2001; Guilley et al., 2004; Diaci et al., 2008). 4.4. Conclusions The present location of many oaks stands in steep areas indicates that these stands have remained in such areas from immemorial times because it was not feasible to fell these trees, and these stands are now very valuable in ecological and landscape terms. However, the results show that with lower values of the physiographic parameter, altitude, slope and distance from the sea, the stand density is greater and the spacing between trees is lower. Therefore the best sites for Q. robur in northwest Spain correspond to thalweg zones or intermediate slope with an important oceanic influence. Moreover, in these zones, the climatic characteristics

combine optimally higher minimum temperature, lower thermal amplitude and higher precipitation. This is related to the growth period of the species and guarantee adequate growth. Because of the particular characteristics of Galician soils, the edaphic properties have less influence on the silvicultural variables. As regards the floristic data, the results obtained indicate that the floristic composition of the undergrowth has little effect on the growth and production of pedunculate oak stands. Given the present silvicultural status of these stands, alternative methods must be proposed for their management, which will range from a conversion to high forest to recovery of the most deteriorated stands by reforestation. Acknowledgements The present study was financed by the Direccio´n Xeral de Investigacio´n e Desenvolvemento of the Consellerı´a de Innovacio´n, Industria e Comercio of the Xunta de Galicia, within research projects PGIDT99MA29101 and PGIDIT02RFO29101PR, carried out in the Department of Agroforestry Engineering, University of Santiago de Compostela. References Aranda, G., 1990. Los bosques flotantes: historia de un roble del siglo XVIII. Collection technique of ICONA, Madrid, Spain, Ministerio de Agricultura, Pesca y Alimentacio´n. Assmann, E., 1970. The Principles of Forest Yield Study. Pergamon Press, Oxford, 506 pp. Aude, E., Lawesson, J.E., 1998. Vegetation in Danish beech forests: the importance of soil, microclimate and management factors, evaluated by variation partitioning. Plant Ecol. 134, 53–65. Aussenac, G., 2000. Interactions between forest stands and microclimate: ecophysiological aspects and consequences for silviculture. Ann. For. Sci. 57, 287–301. Bary-Lenger, A., Nebout, J.P., 1993. Le cheˆne pe´doncule´ et sessile en France et en Belgique. Gerfaut Club Editions du Perron, Alleur-Lie`ge. 604 pp. Berges, L., Gegout, J.C., Franc, A., 2006. Can understory vegetation accurately predict site index? A comparative study using floristic and abiotic indices in sessile oak (Quercus petraea Liebl.) stands in northern France. Ann. For. Sci. 63, 31–42. Bouchon, J., Trencia, J., 1990. Sylviculture et production du cheˆne. Rev. For. Fr. 2, 246–253. Braun-Blanquet, J., 1979. Fitosociologı´a. Base para el estudio de las comunidades vegetales. Ed. Blume. 820 pp. Carballeira, A., Devesa, C., Retuerto, R., Santillan, E., Ucieda, F., 1983. Bioclimatologı´a de Galicia. Xunta de Galicia-Fundacio´n Barrie de la Maza. 391 pp. Decocq, G., Aubert, M., Dupont, F., Alard, D., Saguez, R., Wattez-Franger, A., 2004. Plant diversity in a managed temperate deciduous forest: understorey response to two silvicultural systems. J. Appl. Ecol. 41, 1065–1079. del Rio, S., Penas, A., 2006. Potential distribution of semi-deciduous forests in Castile and Leon (Spain) in relation to climatic variations. Plant Ecol. 185, 269–282. DGCONA, 2001. III Inventario Forestal Nacional, Galicia. Ministerio de Medio Ambiente. 1041 pp. Diaci, J., Gyoerek, N., Gliha, J., Nagel, T.A., 2008. Response of Quercus robur L. seedlings to north–south asymmetry of light within gaps in floodplain forests of Slovenia. Ann. For. Sci. 65 . Dı´az-Maroto, I.J., Vila-Lameiro, P., Silva-Pando, F.J., 2005. Autoecology of oaks (Quercus robur L.) in Galicia (Spain). Ann. For. Sci. 62, 737–749. Dı´az-Maroto, I.J., Vila-Lameiro, P., 2006. Comparison between chemical properties and macronutrients content in natural forests of Quercus robur L. and Quercus pyrenaica Willd. in Galicia, Spain. Agrochimica 50, 200–209. Dı´az-Maroto, I.J., Ferna´ndez-Parajes, J., Vila-Lameiro, P., 2006. Autecology of rebollo oak (Quercus pyrenaica Willd.) in Galicia (Spain). Ann. For. Sci. 63, 157–167. Dı´az-Maroto, I.J., Vila-Lameiro, P., 2007. Deciduous and semi-deciduous oak forests (Quercus robur, Q. petraea and Q. pyrenaica) floristic composition in the Northwest Iberian Peninsula. Biologia 62, 1–10. FAO, 2006. World Reference Base for Soil Resources, 2nd ed., World Soil Resources Reports 103. 142 pp. Gittins, R., 1985. Canonical Analysis. A Review with Applications in Ecology. Springer-Verlag, Berlin, 351 pp. Guchu, E., Dı´az-Maroto, M.C., Dı´az-Maroto, I.J., Vila-Lameiro, P., Pe´rez-Coello, M.S., 2006. Influence of the species and geographical location on volatile composition of Spanish oak wood (Quercus petraea Liebl. and Quercus robur L.). J. Agric. Food Chem. 54, 3062–3066. Guilley, E., Herve, J.C., Nepveu, G., 2004. The influence of site quality, silviculture and region on wood density mixed model in Quercus petraea Liebl. For. Ecol. Manage. 189, 111–121. Harmer, R., Morgan, G., 2007. Development of Quercus robur advance regeneration following canopy reduction in an oak woodland. Forestry 80, 137–149.

I.J. Dı´az-Maroto, P. Vila-Lameiro / Forest Ecology and Management 256 (2008) 702–711 Hummel F.C., Locke G.M., Jeffers J.N., Christie, J.M., 1959. Code of sample plot procedure. Bulletin 31, Forestry Commission. Izco, J., Amigo, J., Guitia´n, J., 1990. Los robledales galaico-septentrionales. Acta Bota´nica Malacitana 15, 267–276. Jongman, R.H.G., ter Braak, C.J.F., Van Tongeren, O.F.R., 1995. Data Analysis in Community and Landscape Ecology. Cambridge University Press, 292 pp. Johnson, R.A., 2005. Probability and Statistics for Engineers, 7th ed. Prentice Hall, Wisconsin, 639 pp. Johnson, R.H., Shifley, R.R., Rogers, R., 2002. The Ecology and Silviculture of Oaks. CAB International, Wallingford, 650 pp. Kelly, D.L., 2002. The regeneration of Quercus petraea (sessile oak) in southwest Ireland: a 25-year experimental study. For. Ecol. Manage. 166, 207–226. Larsen, J.B., 1995. Ecological stability of forest and sustainable silviculture. For. Ecol. Manage. 73, 85–96. Lo¨f, M., Rydberg, D., Bolte, A., 2006. Mounding site preparation for forest restoration: survival and short term growth response in Quercus robur L. seedlings. For. Ecol. Manage. 232, 19–25. Manuel, C.M., Gil, L., 2001. La transformacio´n histo´rica del paisaje forestal en Galicia. Ministerio de Medio Ambiente, Madrid, Spain. Park, A.D., 2001. Environmental influences on post-harvest natural regeneration in Mexican pine-oak forests. For. Ecol. Manage. 144, 213–228. Reque, J., Bravo, F., 2007. Viability of thinning sessile oak stands by girdling. Forestry 80, 193–199. Retuerto, R., Carballeira, A., 1990. Phytoecological importance, mutual redundancy and phytological threshold values of certain climatic factors. Vegetatio 90, 47–62. Retuerto, R., Carballeira, A., 1991. Defining phytoclimatic units in Galicia, Spain, by means multivariate methods. J. Veg. Sci. 2, 669–710. Rivas-Martı´nez, S., Ferna´ndez, F., Loidi, J., Lousa˜, M., Penas, A., 2001. Syntaxonomical checklist of vascular plants communities of Spain and Portugal to association level. Itinera Geobotanica 14, 5–341. Roda´, F., Retana, J., Gracia, C.A., Bellot, J., 1999. Ecology of Mediterranean Evergreen Oak Forests. Springer, Berlin, 373 pp.

711

Rondeux, J., 1993. La mesure des arbres et des peuplements forestiers. Les Presses Agronomiques de Gembloux, Belgique, 541 pp. Rubio, A., Escudero, A., Gandullo, J.M., 1997. Sweet chestnut silviculture in a ecological extreme of its range in the west of Spain (Extremadura). Ann. For. Sci. 54, 667–680. ˜ a. Direccio´n General de ConservaRuiz de la Torre, J., 1991. Mapa Forestal de Espan cio´n de la Naturaleza, Instituto Geogra´fico Nacional, Ministerio de Medio Ambiente. 1870 pp. Russell, J.S., Moore, A.W., 1968. Comparison of different depth weightings in the numerical analysis of anisotropic soil profile data. In: Proceedings of the Ninth International Congress on Soil Science, vol. 4. pp. 205–213. SAS Institute Inc., 2004. SAS/STAT1 9.1. User’s Guide. Cary, SAS Institute Inc. CdRom. Swanson, F.J., Franklin, J.F., 1992. New forestry principles from ecosystem analysis of Pacific Northwest forest. Ecol. Appl. 2, 262–274. Tarrega, R., Calvo, L., Marcos, E., Taboada, A., 2007. Comparison of understory plant community composition and soil characteristics in Quercus pyrenaica stands with different human uses. For. Ecol. Manage. 241, 235–242. ter Braak, C.J.F., 1994. Canonical community ordination. Part I: Basic theory and linear methods. Ecoscience 1, 127–140. Timbal, J., Aussenac, G., 1996. An overview of ecology and silviculture of indigenous oaks in France. Ann. For. Sci. 53, 583–591. Valkonen, S., Urpelainen, P., Virkki, A., 2002. Yield potential of Quercus robur stands in Finland. Scand. J. For. Res. 17, 248–255. Van Calster, H., Baeten, L., De Schrijver, A., De Keersmaeker, L., Rogister, J.E., Verheyen, K., Hermy, M., 2007. Management driven changes (1967–2005) in soil acidity and the understorey plant community following conversion of a coppice-with-standards forest. For. Ecol. Manage. 241, 258–271. Vila-Lameiro, P., Dı´az-Maroto, I.J., 2005. Study of the influence of environmental factors on the width of growth rings in Quercus petraea (Matts.) Liebl. through interpretation of digital images. Wood Res. 50 (3), 23–36. Zaczek, J.J., 2002. Composition, diversity, and height of tree regeneration, 3 years after soil scarification in a mixed-oak shelterwood. For. Ecol. Man. 163, 205–215.