Régénération naturelle de trois essences majeures des peuplements forestiers

.INTRODUCTION

Oak (Quercus petraea (Matt) Liebl., Quercus robur L.), European beech (Fagus sylvatica L.) and hornbeam (Carpinus betulus L.) are key European broadleaved species having distribution areas that largely overlap in Europe (San-Miguel-Ayanz et al., 2016). In the overlapping area, forest managers usually aim to favour oak, which represents greater economic and ecological interest (Thomas et al., 2002; Annighöfer et al., 2015). The prospects for climate change impel towards a more oak-oriented forestry. Indeed, beech and hornbeam being more sensitive to drought than oak (Bréda et al., 2006; Friedrichs et al., 2009), their future remain uncertain in the context of more intense and/or more frequent spring and summer droughts (Jump et al., 2006; Geßler et al., 2007). However, beech and hornbeam are strong competitors for oak, especially during the regeneration stage (Ligot et al., 2013; Petritan et al., 2012; Van Couwenberghe et al., 2013) and have strong negative impacts on the survival, height and diameter growth of young oaks (Jensen & Löf, 2017; De Groote et al., 2018; Maleki et al., 2020). In order to maintain or to promote oak in European forests of the next century, it is crucial to identify which environmental and silvicultural conditions would favour the regeneration of oak compared to that of hornbeam and beech. Distance to seed source is a major determinant of regeneration success, for all tree species (Clark et al., 1999; Nathan & Muller-Landau, 2000), and has been shown to predominate over competition from tree seedlings or neighbouring vegetation in mixed forests of north-eastern France (Dassot and Collet 2015). Large distance from potential seed sources reduces seedling density, according to patterns that may differ among tree species (Clark et al., 1998; Hewitt & Kellman, 2002; Rozman et al., 2015). The dissemination strategies of oak, beech and hornbeam differ (Vittoz & Engler, 2007): beech and oak are barochory while hornbeam is anemochore and usually disseminates over longer distances (Vittoz & Engler, 2007). Several studies have shown the impact of surrounding stand composition (Petritan et al., 2012; Annighöfer et al., 2015; Manso et al., 2020) or of distance to the gap edge on the density of oak and beech saplings in canopy gaps (Degen, 2006; Van Couwenberghe et al., 2010; Tinya et al., 2020). To date, very few studies (Tinya et al., 2020) have quantified these effects for hornbeam.

MATERIALS & METHODS

Study sites

The study was conducted in temperate forests in France (Figure 20). Selected forests were mainly semi-natural hardwood stands dominated by oak, beech and hornbeam, located in lowlands areas, with an altitude below 600 m. Mean annual temperature ranges between 7.9 °C and 11.85 °C and annual precipitation ranges from 593 mm to 1505 mm. Figure 20 | Location of the 108 study sites in the areas affected by the windstorms of 1999 in France. In December 1999, 968 000 ha of French forests were affected by two windstorms, Lothar and Martin (Inventaire Forestier National, 2003), which destroyed 8.3 % of French forest resource (Pignard et al., 2009). In 2001, 108 study sites were established in forests impacted by Chapitre 4 – Régénération naturelle de trois essences majeures des peuplements forestiers français et européens -70- the windstorms (Figure 20). We selected study sites in order to investigate large gradients of disturbance intensity (damaged area ranged between 10 % and 90 % of the forest surface area), soil conditions (pH and C/N of the organo-mineral horizon A ranged from 3.9 to 7.9 and 11 to 28 respectively) and previous stand type (oak-dominated, beech-dominated, or mixed oak and beech) (Tableau 14). One to three gaps were selected in each site. The area of selected gaps ranged from 0.02 to 144 ha. In total 108 gaps were selected for the study. No silvicultural operations were performed in the selected gaps after the storms.

Data collection

Gaps were classified according to their surface area with a threshold of 1 ha to differentiate small gaps (SG) from large gaps (LG). Two different sampling designs were used for measurements in SG and LG (Supplementary, Figure 25). In SG, 2-m-radius circular plots (12.6 m2 ) were distributed every 12 m, away from the barycentre of the gaps, along the northsouth and east-west axes of the gap. The axes extended outside the gap and one or two plots were placed under the forest cover. The number of plots differed among gaps, and ranged between 7 and 24 plots, depending on gap size. In LG, the centre of the gap was approximately located, and four 10-m-radius circles were established at 50 gr, 150 gr, 250 gr and 350 gr (0 gr indicating north) and 30 m with respect of the gap centre. In each circle, three 2-m-radius circular plots were established at 0 gr, 133 gr and 266 gr respectively with respect to the circle centre. Twelve plots were installed in each LG. A total of 1153 plots of 12.6 m2 (592 in small gaps, 561 in large gaps) were established between 2001 and 2004 for the study. An agreement with forest managers prohibited any anthropic intervention, allowing the spontaneous dynamic to occur. In spring and summer 2018 and 2019, (i.e. 19 and 20 growing seasons after the windstorms occurred), tree regeneration measurements were performed in each plot. All trees with a height above 0.1 m were considered. Trees with a diameter greater than 20 cm were considered as preexistent to the storm and were not considered. The species was recorded for each tree. Diameter at breast height was measured on each tree with a height above 1.3 m. Within each plot and each species, the height at the beginning of the growing season was measured for the tree with the highest diameter and the tree with the median diameter. A total of 19445 saplings were recorded, 7341 and 1756 with a diameter and a height measurement, respectively (Tableau 15).