Effect of Site Quality and Thinning Management on the Structure of Holm Oak Forests in Northeast Spain

Original article

Effect of site quality and thinning management on the structure of holm oak forests in northeast Spain

M Gracia J Retana

1 Centre de Recerca Ecològica i Aplicacions Forestals, Facultat de Ciències, Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona; 2 Departament de Producciò Vegetal i Ciència Forestal, Universitat de Lleida, Rovira Roure 177, 25006 Lleida, Spain

(Received 6 September 1994; accepted 20 November 1994)

Summary — In the holm oak forests of NE Spain, the variability of thinning intensity, together with the great heterogeneity of site quality, due to the rough topography of the region, gives rise to important structural differences among stands. In this paper, we analyze the effect of site quality and thinning management on the structural heterogeneity of holm oak stands at the end of the cutting cycle in different areas of the Montseny and Les Guilleries massifs (NE Spain). These two factors influence the main structural variables of the stands in various ways. Height was used as an indicator of site quality, because it was relatively unaffected by thinning management, while density and total number of stems per stool were strongly affected. Mean tree diameter reflected the effect of both site quality and thinning. The type of forest management was found to be the most important factor determining forest structure. Site quality was a constraint of stand development, contributing in poor stands, to increased differences due to thinning management. holm oak / Ouercus ilex / forest structure / site quality / thinning

Résumé — Effets de la fertilité de la station et de l’intensité d’éclaircie sur la structure des forêts de chêne vert du nord-est de l’Espagne. Dans les forêts de chêne vert du nord-est de l’Es- pagne, la variabilité de l’intensité d’éclaircie et la grande hétérogénéité de fertilité des stations, dues à leur topographie montagneuse, provoquent d’importantes différences de structure des peuplements. Dans cet article, nous analysons l’effet de la fertilité de la station et de la gestion sylvicole sur l’hété- rogénéité structurelle des forêts de chêne vert dans différentes zones des massifs de Montseny et les Guilleries (nord-est de l’Espagne). Ces deux facteurs influencent de différentes manières les principales variables descriptives de la structure des peuplements. La hauteur a été utilisée comme indice de fertilité de station car elle est très peu modifiée par le type de gestion sylvicole, tandis que la densité et le nombre total de brins par souche sont très affectés par le type de gestion. La diamètre moyen syn- thétise l’effet de la fertilité de la station et de l’intensité d’éclaircie. La gestion sylvicole est le facteur le plus important pour expliquer les différences de structure de peuplement, tandis que la fertilité de la station peut être un facteur limitant du développement des peuplements, contribuant, dans les stations de moindre qualité, à augmenter les différences dues à la gestion sylvicole. chêne vert / Quercus ilex / structure de la forét / fertilité de la station / éclaircie

INTRODUCTION Castri, 1973; Terradas, 1991; Barbero et al, 1992). Heterogeneity is an intrinsic feature of the In this study we analyze the effect of site Mediterranean landscape (Naveh and quality and thinning management on the Lieberman, 1984; Barbero, 1988). This het- structural heterogeneity of holm oak forests erogeneity appears not only in the variation at the end of the cutting cycle. The initial of holm oak forests to different of species composition associated with gra- response fertilization and dients of aspect, elevation or soil depth thinning, irrigation practices (Aschmann, 1973; Rabinovitch-Vin, 1983; has been previously studied by measuring Quezel and Barbero, 1989; Pigott and Pig- parameters such as growth of remaining ott, 1993), but also in the changes of stand trees, production and dynamics of sprouts, structure and architecture of trees observed etc (Cartan-Son et al, 1992; Ducrey, 1992; within the same stand. It is therefore nec- Ducrey and Turrel, 1992; Floret et al, 1992; and essary to carry out a description of these Mayor Rodá, 1994). Nevertheless, these studies are and it forests based on structural, architectural relatively recent, has not been to evaluate the and physiognomic features on the different yet possible relative of these factors in the vertical strata of the canopy, that provide importance information about this dynamic environment, structural features of the stands in the medium term. which is a habitat of animal communities (Cody, 1985; De Garnica and Robles, 1991), liable to suffer disturbances (Sousa, 1984; STUDY AREA AND METHODS Lorimer, 1989; Oliver and Larson, 1990), and potentially be used and exploited in numerous ways (Ducrey, 1992). This study was carried out in three different areas: Gualba, in the southwestern sector of the Holm oak forests (Quercus ilex) coppice Montseny Massif (40 km from Barcelona, Spain), are one of the most representative com- Arbúcies, in the northern sector of the Montseny munities in the western part of the Mediter- Massif (90 km from Barcelona), and Susqueda, in ranean basin (they cover more than 120 000 the southwestern sector of the Les Guilleries Mas- ha in Catalonia alone; Montoya, 1988). In sif (110 km from Barcelona). The climate of the is subhumid Mediterranean. Mean annual this type of forest, the inherent heterogeneity region from 600 to 800 mm and of Mediterranean has precipitations range ecosystems multiple mean annual temperatures from 10 to 15 °C. origins that overlap on different temporal Soils are developed on a bedrock of granite and and spatial scales (David et al, 1979; Floret metamorphic schists. The main vegetation type of et al, 1989; Barbero, 1988; Romane et al, the area corresponds to the montane holm-oak 1988). In addition to this environmental het- forest (Quercetum mediterraneo-montanum), with Q ilex as the main at the tree there is one another caused component layer, erogeneity, by and Erica arborea and Arbutus unedo as the main human as this of forest influence, type components of the shrub layer. Holm oak forests ecosystem has been subjected for decades of this region are managed by coppicing through or even centuries to frequent coppicing (Di selective cutting at intervals of approximately 30 years (Retana et al, 1992). Together with this ning intensity. In each of the tree areas, groups of variability in management practices, there is a plots of similar site quality (plots which did not spatial heterogeneity in site quality due to the show significant differences in height from an rough topography of the region. To distinguish analysis of variance) have been established. A between the effects of management and quality, group of low quality plots and another of high plots which differed in only one of these two fac- quality plots have been distinguished in each area. tors have been compared. Circular plots with a 12 m radius (450 m2) were In each area, two adjacent stands with differ- established. In each plot, the following measures ent thinning management were chosen: one qual- were taken: ified as ’conservative and another as thinning’, — topographic factors: elevation, aspect, slope ’intense In the area of a third thinning’. Susqueda, and position on the slope (low, medium and high) stand, referred to as ’old forest’ was also included. The three stands corresponding to conservative — diameter at breast height (dbh) of all standing trees than 5 cm dbh thinning belonged to the same owner, and were larger managed by selective thinning. This practice — height of the ten tallest trees in the plot involved a period of about 25-30 years between Forest structure has been described using the harvests, a variable harvest of basal area following variables: (40-60%), and special care in the selection of — of stems above 5 cm dbh the remaining trees. These stands had a similar density age from last thinning. The three stands corre- — basal area to were intense all character- — sponding thinning mean dbh of stems above 5 cm dbh ized by a less careful management. The high cutting intensities of these stands had the — dominant height characteristics of clear felling practice in some — number of stems per stool cases. a They presented longer period between — stool equivalent diameter, which is the diameter were con- harvests than stands which thinned that has the same basal area as the sum of the The stand of old forest in servatively. Susqueda basal areas of all stems from the stool. was a portion of the conservatively thinned stand To elaborate on the diameter distributions, that was not cut in the last thinning period. There- stems of each were in width intervals fore, its age from last thinning was 70 years plot grouped instead of 35. of 2.5 cm, starting from the initial lowest value of 5 cm dbh. Diameter distributions have been com- In each stand, and within the limits of the same pared by means of the Kolmogorov-Smirnov test. slope, several plots which seemed to be representative of the best and the worst qualities were chosen. The selection of plots was based on an assessment of the general aspect of the plot and RESULTS the size and physiognomy of holm oaks. Table I summarizes the main features of the different plots considered in the present study. To evaluate Overall analysis of the three stands the effect of site on forest quality structure, plots of conservative thinning of different stands with the same thinning management (including criteria of stem selection, intensity and age from last thinning), have been In order to analyze features in stands with To evaluate the effect of man- compared. thinning similar thinning management, the nine plots agement, plots of different stands of the same of the three stands with conservative thin- area with similar quality were compared. Domi- nant height has been chosen as site index, based ning have been considered together. Mean on data obtained in a previous study (Gracia, in values of structural features for these plots preparation), because it attains several objec- are summarized in table II. Differences tives: it shows with 1) significant relationships among plots were significant for all struc- site it is the structural variable that is variables; 2) tural variables P most closely correlated with other structural vari- (F-test, < 0.0001), although the Fisher LSD ables, especially mean diameter, when man- multiple-comparison proce- agement is constant; and 3) it stabilizes very dure (Ott, 1988) showed different trends: a) quickly after thinning, and is not affected by thin- mean height of different plots had a continuous gradient, which makes it impos- other variable, especially with the mean sible to define independent groups, except diameter (r = 0.98), but also with the total for the one composed by plot cA4, of high- density of stems (r = -0.89) and the stool est quality; b) three different groups of plots equivalent diameter (r = 0.88). The rela- were identified according to mean diame- tionship between height and the number of ter; once again the cA4 plot formed an inde- stems per stool was lower (r=-0.62), while pendent group of exceptionally large mean the relationship with basal area was not sig- diameter; c) the results concerning the num- nificant. The mean diameter also displayed ber of stems per stool were even more sim- accurate correlation coefficients with the plified, because only two independent other variables, even with basal area, which groups of plots could be identified: those was not significantly correlated with the other with a large number of stems per stool, structural variables in this analysis. which also had the smallest mean diameters and mean heights, and those with a small number of stems per stool, which also Comparison of different stands had the largest mean diameters and mean within the same area heights; and d) it was not possible to identify different for the stool groups equivalent In each of the groups of plots with similar diameter because the values of different site quality, we have compared the differ- were those of plots quite similar, excepting ent variables considered. In the areas of the plots of highest quality (cA4, cG3). Arbúcies (table IV) and Gualba (table V), A correlation analysis between the main the plots of stands managed by conservative structural variables of the nine plots with thinning and those managed by intense thin- conservative thinning has been carried out ning were clearly distinguished in both low (table III). Height was correlated with most and high quality sites. Plots from stands under intense thinning had smaller mean ity did not show significant differences in diameters and a greater number of stems mean diameter, and only differed in the per stool than plots from stands under by mean number of stems per stool between conservative thinning. In the area of the stand of intense thinning and the stands Susqueda (table VI), plots of similar qual- of conservative thinning and old forest. The stool equivalent diameter, in general, did related with the other structural variables, not vary in plots of similar quality and dif- showing similar values to those obtained ferent type of management, excepting those with only the plots under conservative thin- of high quality in Gualba and Susqueda ning (table III). Basal area alone presented (tables IV and VI). a significant, but low, correlation with height (r= 0.47).

Relationship among the structural variables considered Diameter distribution in the different plots Another correlation analysis has been carried out with all plots of different thinning Within the same stand, the low quality plots intensity (table VII). In general, correlation displayed a greater number of small stems values decreased with respect to those than the high quality plots. The shape of the obtained with only the plots of conservative distribution changed severely. The differ- thinning (table III). Height maintained good ences between the extreme situations were correlations with the stool equivalent diam- significant in all stands (Kolmogorov- eter (r = 0.81) and the mean diameter (r= Smirnov test, P < 0.10). This trend is shown 0.76), but not with the total density of stems in figure 1, where the diameter distributions or the number of stems per stool. The mean of the four plots of stand cA are in ascend- diameter was the variable most closely cor- ing order from cA1 to cA4.

The comparison of plots of similar site distributions in plots under conservative thin- quality and different management showed ning and old thinning of Susqueda (fig 4) several differences: a) in high quality plots were not significantly different (Kol- (fig 2), differences due to management were mogorov-Smirnov test, P > 0.10), although only significant in the area of Gualba (Kol- there were fewer intermediate-sized stems mogorov-Smirnov test, P < 0.05), but not in the more recently cut plots. in the other two areas (Kolmogorov-Smirnov test, P > 0.10); and b) in low quality plots, DISCUSSION there were significant differences among plots of the two areas due to management (fig 3) (Kolmogorov-Smirnov test, P < 0.05): Effect of site and conservative management led to an almost quality thinning on the structural variables of the holm normal distribution, while intense manage- and stands ment led to a unimodel distribution. The effect of thinning even reversed the usual trend of a decreasing number of small stems The changes in site quality and manage- with increasing site quality: plot cS2 pre- ment have different effects on the structural sented a distribution closer to normal than variables of the stands: several variables, the higher quality plot iS2 (fig 3). Diameter such as height, can be considered as plot characteristics and, within certain limits, only densities, as would a decrease in site qual- depending on site quality (Gracia, in prepa- ity. The number of stems per stool is also ration); nevertheless, the majority of vari- strongly affected by management, as can ables are more or less affected both by site be deduced from the comparison of stands quality and by management. Density is within the same area. It is less affected by related to site quality, as can be deduced site quality than density and can thus be from the correlation obtained between the considered a more suitable variable for dis- plots with conservative thinning. This result criminating between different thinning man- agrees with the one given by Lledó et al agements. (1992) for a holm oak forest in Prades (Tar- Data obtained in this study (ie, the high ragona), where the differences in density correlation between height and diameter for found between the low and high elevation the plots with conservative thinning, which zones of the same slope have been decreases when all plots of different man- attributed to differences in productivity (ie, agement are considered) support the strong site quality). This close relationship between dependence with diameter on both site qual- height and density disappears when all plots ity and managements described in other are taken into account, because of the over- studies (Cartan-Son et al, 1992; Ducrey and siding effect of management on density. A Turrel, 1992; Mayor and Rodá, 1993). That high thinning intensity would lead to low is why this variable is probably the one that yields the most information about forest the silvicultural point of view (Montero, structure. Nevertheless, and for the same 1992). Although basal area is the variable most related with forest biomass reason, it does not allow, on its own, an closely its analysis of the role played by different fac- (Bengoa, 1993), extrapolation to productivity is quite difficult, because it would be tors on the actual forest structure. necessary to include a time factor and to Nothing has been said about basal area, know, for selective thinning, the amount of one of the most important variables from timber harvested. In the studied plots, basal area was a very unpredictable variable, one-stemmed stools of large size, with low which only showed a slight tendency to density and a tendency towards uniformity. increase with site quality. The similarity between the structural features of the studied plots and this ideal, mature state allows us to discuss the role Site quality and thinning management of site quality and management as factors as factors conditioning the structural or the development of holm oak forest slowing down, stabilizing accelerating developmental process of holm oak forests. When other factors are similar, site charac- (1992) proposed a theoretical cycle Ducrey teristics move this development towards of development of holm oak forests, with a mature forest. In high quality plots, forest coppice stabilized by clear felling practices, structure is similar to that of a theoretically and a maturity stage that would be reached mature forest. This is evident in after a long period of time without distur- particularly the old stand. Lledo et al bances. This idea of a structural maturity of thinning (1992) holm oak forests is also used by other also observed in holm oak forests in Prades authors (Barbero, 1988; Lledó et al, 1992), (Tarragona) that high quality areas show a and could be defined as a forest formed by greater degree of maturity than low quality areas, with smaller densities and greater responsible for the heterogeneity of the stud- mean diameters. ied holm oak forests. Given the range of This pattern is also clearly observed in variation among these factors in the region, has a the diameter distributions. For a given stand, management probably larger importance in this Site either the lower quality plots show a greater num- heterogeneity. quality ber of small diameter stems than those of favors or limits forest development, as sug- higher quality. Courraud (1987) explains gested by the greater differences between this fact by considering that in low quality types of management found in low quality than in In extreme sit- sites very few trees attain superiority in size plots high quality plots. in relation to their neighbors sufficient to uations with incomplete cover, we could find exclude them. In high quality sites, there is what Barbero (1988) defined as the pro- a decrease in individuals in the smaller gressive architectural model, characterized classes, with a tendency to normality which by stools with a great number of stems is clearest in the old forest stand (oS). Bar- which coexist from the subhorizontal to the bero (1988) describes this pattern in vertical position. However, in the study area, monospecific holm oak forests with com- this pattern is only found in very disturbed plete cover by means of a multistrata struc- forests or in very rocky areas. The results tural model, characterized by having sev- presented in this paper suggest the role of eral dominant sprouts per stool and a site quality and thinning management on greater number of suppressed sprouts, the structure of holm oak forests. A deeper which, in high quality sites, are very quickly knowledge of these topics will make it pos- eliminated. sible to provide basic information for a more coherent forest management In the studied plots, managements is an design. essential factor affecting structural development. The old plots are obviously the clos- ACKNOWLEDGMENTS est to our ideal of a mature forest and particularly the high quality plot has a high forest physiognomy, with one-stemmed stools, a Thanks are due to E Garolera and P Tortadés monospecific composition in the overstory, for allowing us to carry out this study on their and to T for cor- and an almost completely absent under- properties, Lokingbill English rection. This research was funded INIA The stands under conservative thin- partly by story. project SC94-023. ning show values of the structural variables similar to those obtained for the old thinning stand. It is quite surprising how quickly the REFERENCES holm oak forest recovers from thinning, which that this of suggests type manage- Aschmann H (1973) Man’s impact on the several regions ment stabilizes the forest in a state very with mediterranean climates. In: Mediterranean-type similar to the one described as mature for- Ecosystems. Origin and Structure (F Di Castri, HA est. In contrast, the plots exploited by Mooney, eds), Springer-Verlag, Berlin, 363-372 Barbero M Caractérisation de struc- intense show differences com- (1988) quelques thinning large tures et architectures forestières des arbres et with the mature forest model with pared arbustes à feuilles persistantes de l’étage méditer- respect to the number of stems per stool, ranéen. Rev For Fr 40, 371-380 density, multispecific composition of the Barbero M, Loisel R, Quezel P (1992) Biogeography, overstory and degree of development of the ecology and history of Mediterranean Quercus ilex understory. ecosystems. Vegetatio 99-100, 19-34 Bengoa J (1993) Calidad de estación en rebollares de la The results obtained suggest that both Rioja: altemativas al Indice de Sitio. I Congreso Fore- site quality and thinning management are stal Español, vol II, 473-478 Cartan-Son M, Floret C, Galan MJ et al (1992) Factors Montseny mountains (NE Spain). Ann Sci For 50, affecting radial growth of Quercus ilex in a coppice 247-256 stand in southern France. 99-100, 61-68 Vegetatio Mayor X, Rodá F (1994) Effects of irrigation and fertil- Cody ML (1985) An introduction to habitat selection in ization on oak diameter growth in a Mediterranean birds. In: Habitat Selection in Birds (ML Cody, ed), holm oak forest. For Ecol Manage 68, 119-126 Academic Press, London, 3-56 Montero G (1992) Aspectos ecológicos y productivos Courraud R (1987) Diagnostic des peuplements de de la selvicultura. Ecologia 6, 111-121 chênes. Forêt-Entreprise 45, 96-111 Montoya JM (1988) Encinas y encinares. Ediciones David F, Poissonet P, Romane F (1979) Analyse de la Mundi-Prensa, Madrid structure horizontale d’un taillis de Bouleaux (Betula Naveh Lleberman AS The- verrucosa Ehrh) en Sologne. Acta Oecol, Oecol Plant Z, (1984) Lanscape Ecology: 14, 237-247 ory and Application. Springer-Verlag, Berlin De Garnica R, Robles L (1991) Un método para la clasi- Oliver CD, Larson BC (1990) Forest Stand Dynamics. ficiación fisionómica de las encinas y los encinares. McGraw Hill, New York Ecología 5, 173-180 Ott L (1988) An Introduction to Statistical Methods and Di Castri F (1973) Soil animals in latitudinal and topo- Data Analysis. PWS-Kent Publishing Company, graphical gradients of mediterranean ecosystems. Boston In: and Mediterranean-type Ecosystems. Origin Pigott CD, Pigott S (1993) Water as a determinant of Structure Di Castri, HA (F Mooney, eds), Springer the distribution of trees at the boundary of the Verlag, Berlin, 171-190 Mediterranean zone. J Ecol 81, 557-566 Ducrey M (1992) Quelle sylviculture et quel avenir pour Quezel P, Barbero M (1989) Zonation altitudinale des les taillis de chêne vert (Quercus ilex L) de la région structures forestiéres de en California méditerranéenne. Rev For Fr 44, 12-34 végétation méditerranéenne. Leur interprétation en fonction des Turrel M Influence of methods Ducrey M, (1992) cutting méthodes utilisées sur le pourtour méditerranéen. and dates on stump sprouting in holm oak (Quer- Ann Sci For 46, 233-250 cus ilex L) coppice. Ann Sci For 49, 449-464 Rabinovitch-Vin A (1983) Influence of nutrients on the Floret Galan Le Floc’h Romane F C, MJ, E, Rapp M, composition and distribution of plant communities de la de la biomasse (1989) Organisation structure, in mediterranean-type ecosystems of Israel. In: et de la minéralomasse d’un taillis ouvert de chêne Mediterranean-type Ecosystems. The Role of Nutri- vert ilex Acta Oecol, Oecol Plant 10, (Quercus L). ents DT JUM 245-262 (FS Kruger, Mitchell, Jarvis, eds), Springer-Verlag, Berlin, 74-85 Floret C, Galan MJ, Le Floc’h E, Romane F (1992) Retana Riba Castell JM Dynamics of holm oak (Quercus ilex) coppices after J, M, C, Espelta (1992) Regen- of clearcutting in southern France. Flora and life cycle eration by sprouting holm-oak (Quercus ilex) changes. Vegetatio 99-100, 97-106 stands exploited by selection thinning. Vegetatio 99- 100, 355-364 Lledó MJ, Sánchez JR, Bellot J, Boronat J, ibáñez JJ, Escarré A (1992) Structure, biomass and produc- Romane F, Floret C, Galan M, Grandjanny M, Le Floc’h tion of a resprouted holm oak (Quercus ilex) forest in E, Maistro M, Perret M (1988) Quelques remarques NE Spain. Vegetatio 99-100, 69-76 sur les taillis de chênes verts. Répartition, histoire, biomasse. Forêt Méditerranéenne 10, 131-135 Lorimer CG (1989) Relative effects of small and large disturbances on temperate hardwood forest structure. Sousa WP (1984) The role of disturbance in natural Ecology 70, 565-567 communities. Ann Rev Ecol Syst 15, 353-391 Mayor X, Rodá F (1993) Growth response of holm oak Terradas J (1991) Altérations du paysage méditerranéen (Quercus ilex) to commercial thinning in the par l’homme. IMCOM 3, 18-22