Survival and growth of upland oak and co-occurring competitor seedlings following single and repeated prescribed fires

Studies within and outside the U.S. indicate recurring oak (Quercus spp.) regeneration problems. In deciduous forests of the eastern U.S., a prevailing explanation for this trend is fire suppression leading to high competitor abundance and low understory light. In response, prescribed fire is increasingly used as a management tool to remedy these conditions and encourage future oak establishment and growth. Within eastern Kentucky, we implemented single and repeated (3×) prescribed fires over a 6-yr period (2002–2007). Pre- and post-burn, we quantified canopy cover and oak seedling survival and growth compared to other woody seedlings deemed potential competitors, primarily red maple (Acer rubrum L.) and sassafras (Sassafras albidum (Nutt.) Nees.). Burning temporarily decreased canopy cover 3–10%, but cover rebounded the subsequent growing season. Repeated burning ultimately produced canopy cover about 6% lower than sites unburned and burned once, suggesting a cumulative effect on understory light. Red maple exhibited low survival (∼40%) following single and repeated burns, but growth remained similar to unburned seedlings. Burning had little impact on sassafras survival and led to total height and basal diameters 2× greater than unburned seedlings. A single burn had no impact on red oak (Erythrobalanus spp.) survival and increased height and basal diameters 25–30%, but this positive growth response was driven by seedlings on several plots which experienced high burn temperatures and consequently high overstory mortality. White oaks (Leucobalanus spp.), however, exhibited twice as high mortality compared to those unburned, with no change in growth parameters. Repeated burning negatively impacted survival and growth of both oak groups compared to unburned seedlings. With both burn regimes, oaks with smaller pre-burn basal diameters exhibited the lowest post-burn survival. Thus, despite the ability of prescribed burns to temporarily increase understory light and reduce red maple survival, neither single or repeated burns placed oaks in an improved competitive position. These findings result from a combination of highly variable yet interdependent factors including the (1) life history traits of oaks compared to their co-occurring competitors, (2) pre-burn stature of pre-existing oak seedlings, and (3) variability in fire temperature and effects on understory light.     

Fire-driven oak regeneration in French Mediterranean ecosystems

Regeneration by seeds for cork oak (Quercus suber) and companion oaks (holm oak Quercus ilex and downy oak Quercus pubescens) is likely to be poor in the fire-prone Maures massif (southern France) but the causes are poorly known. Our objective was to assess the effective recruitment for these three oak species and their temporal pattern of recruitment, in order to determine the main limitation factors and the regeneration window of each species. We studied oak recruits (height <3 m) in naturally regenerated populations according to a gradient of fire recurrence and in five main vegetation types including shrublands and mixed mature woodlands. Fire recurrence was the main explanatory factor of oak recruitment, either directly or through vegetation type and microsite characteristics. The results indicate nil to low recruitment for holm oak and downy oak in shrublands, especially those recurrently burned and dominated by Cistus species. Cork oak recruited better than the other oaks in medium and high shrublands dominated by Erica arborea. In contrast, recruitment was high for holm and downy oak in mixed oak stands and mixed pine-oak stands that have not burned for decades. Microsite conditions such as coverage by litter and shrubs influenced oak recruitment, whereas landscape configuration and stand basal area had no influence. Our results suggest that strategic shrub-clearing, oak planting and protection of mixed oak woodlands as seed sources would help maintaining oak populations in the woodland–shrubland mosaic.     

Restoration of bottomland hardwood forests across a treatment intensity gradient

Large-scale restoration of bottomland hardwood forests in the Lower Mississippi Alluvial Valley (USA) under federal incentive programs, begun in the 1990s, initially achieved mixed results. We report here on a comparison of four restoration techniques in terms of survival, accretion of vertical structure, and woody species diversity. The range of treatment intensity allowed us to compare native recolonization to direct seeding and planting of Quercus nuttallii Palmer, and to an intensive treatment of interplanting two species that differed in successional status (early successional Populus deltoides Bartram ex Marsh. ssp. deltoides, with the mid-successional Q. nuttallii). Native recolonization varied in effectiveness by block but overall provided few woody plants. All active restoration methods (planting and direct seeding) were successful in terms of stocking. Populus grew larger than Quercus, reaching canopy closure after 2 years and heights after 2 and 5 years of 6 and 12.7 m, respectively. Planted Quercus were significantly larger than direct seeded Quercus in all years, but only averaged 1.4 m in height after 5 years. Interplanting did not seem to facilitate development of the Quercus seedlings. The early success of the interplanting technique demonstrated that environmental benefits can be obtained quickly by more intensive efforts. Native recolonization can augment active interventions if limitations to dispersal distance are recognized. These results should provide landowners and managers with the confidence to use techniques of varying intensity to restore ecosystem functions.     

Ecological consequences of an exotic fungal disease in eastern U.S. hardwood forests

Exotic pests and pathogens can cause extensive mortality of native species resulting in cascading effects within an ecosystem. As ecosystems lose species to exotic enemies, ecosystem function may be disrupted if the ecological roles are not filled by the remaining species. To illustrate this concept, this paper examines the impacts of an exotic fungus (Discula dectructiva) on flowering dogwood (Cornus florida), historically a common understory tree species in eastern U.S. hardwood forests. Recent studies indicate that dogwood plays an important role in the health and ecological integrity of forest ecosystems throughout the eastern U.S. by increasing the availability of calcium in the biota-rich surface horizons of forest soils. However, Discula destructiva causes a disease, dogwood anthracnose, which can rapidly kill dogwood trees. This paper also illustrates how past fire has increased dogwood density and improved tree health in areas infected with anthracnose, suggesting that prescribed fire may offer a tool for land managers to maintain dogwood as a component in eastern U.S. hardwood forests by shifting the “ideal” disturbance regime of this previously fire-intolerant species.     

Short- and long-term effects of thinning and prescribed fire on carbon stocks in ponderosa pine stands in northern Arizona

Euro-American logging practices, intensive grazing, and fire suppression have increased the amount of carbon that is stored in ponderosa pine (Pinus ponderosa Dougl. Ex Laws) forests in the southwestern United States. Current stand conditions leave these forests prone to high-intensity wildfire, which releases a pulse of carbon emissions and shifts carbon storage from live trees to standing dead trees and woody debris. Thinning and prescribed burning are commonly used to reduce the risk of intense wildfire, but also reduce on-site carbon stocks and release carbon to the atmosphere. This study quantified the impact of thinning on the carbon budgets of five ponderosa pine stands in northern Arizona, including the fossil fuels consumed during logging operations. We used the pre- and post-treatment data on carbon stocks and the Fire and Fuels Extension to the Forest Vegetation Simulator (FEE-FVS) to simulate the long-term effects of intense wildfire, thinning, and repeated prescribed burning on stand carbon storage.The mean total pre-treatment carbon stock, including above-ground live and dead trees, below-ground live and dead trees, and surface fuels across five sites was 74.58 Mg C ha⁻¹ and the post-treatment mean was 50.65 Mg C ha⁻¹ in the first post-treatment year. The mean total carbon release from slash burning, fossil fuels, and logs removed was 21.92 Mg C ha⁻¹. FEE-FVS simulations showed that thinning increased the mean canopy base height, decreased the mean crown bulk density, and increased the mean crowning index, and thus reduced the risk of high-intensity wildfire at all sites. Untreated stands that incurred wildfire once within the next 100 years or once within the next 50 years had greater mean net carbon storage after 100 years compared to treated stands that experienced prescribed fire every 10 years or every 20 years. Treated stands released greater amounts of carbon overall due to repeated prescribed fires, slash burning, and 100% of harvested logs being counted as carbon emissions because they were used for short-lived products. However, after 100 years treated stands stored more carbon in live trees and less carbon in dead trees and surface fuels than untreated stands burned by intense wildfire. The long-term net carbon storage of treated stands was similar or greater than untreated wildfire-burned stands only when a distinction was made between carbon stored in live and dead trees, carbon in logs was stored in long-lived products, and energy in logging slash substituted for fossil fuels.     

Small mammal responses to forest management for oak regeneration in southern Indiana

We used a 5-decade chronosequence of harvest openings to characterize population and community-level responses of small mammals to forest management targeting oak regeneration in southern Indiana. Live-trapping at 42 different sites allowed modeling of occupancy and relative abundance using environmental covariates while incorporating imperfect detection. Species richness was higher in smaller openings on southwest-facing aspects. Similarity between species richness of different age classes decreased with increasing site age. Eastern chipmunk (Tamias striatus) relative abundance was greater in early seral stages, i.e., at young sites with low basal areas. Relative abundance of white-footed mice (Peromyscus leucopus) exhibited different responses to coarse woody debris on sites versus microsites. Pine voles (Microtus pinetorum) and short-tailed shrews (Blarina brevicauda) were more likely to occupy older sites. We observed a greater relative abundance of short-tailed shrews at sites with steep and northeast-facing slopes. Northeast-facing slopes also resulted in higher short-tailed shrew occupancy rates. Incorporating detection probability enabled us to derive more accurate estimates of relative abundance and, when coupled with a Bayesian framework, permitted the estimation of occupancy for uncommon species. Our estimated responses can be used by forest managers to determine the potential impacts of even-aged and uneven-aged oak management on small mammals, and the statistical methodology we used can be applied even more broadly to improve understanding of wildlife responses to forest management.     

Seasonal effects of prescribed burning and roller chopping on saw palmetto in flatwoods

Shrub encroachment has become a problem in many rangeland systems across the United States due to a reduction in the disturbances, primarily fire, which historically maintained them. The shrub saw palmetto (Serenoa repens) has become abundant in many habitats of the southeastern Coastal Plain, including Florida. When fire regimes are altered or fires are suppressed, this species can proliferate leading to significant changes in the ecosystem, particularly the herbaceous vegetation. Prescribed burning and roller chopping are management activities often used to control saw palmetto. However, little is known about the effects these treatments have on this shrub, particularly when applied in different seasons. We compared the seasonal effects of prescribed burning, roller chopping, and combinations of the two on saw palmetto. The effects of treatments on saw palmetto were assessed using a paired-sample approach, where saw palmetto height, cover, and density were compared between sampling locations randomly located within treated (e.g., burned) and untreated areas. Dormant season burning had no effect on saw palmetto density and height and only temporarily reduced cover, with rapid regrowth occurring the first year post-treatment. Growing season burning also had no effect on saw palmetto density. However, saw palmetto cover was lower on growing season burn compared to control sites the first year post-treatment and height the first and second year post-treatment. The combination of burning and roller chopping, despite having no effect on saw palmetto density, did result in lower saw palmetto height compared to controls the first and second year post-treatment. The effect of roller chopping/burning on saw palmetto cover depended on season and year. Saw palmetto cover and height were lower on dormant and growing season roller chop than control sites the first and second year post-treatment, but only growing season roller chopping had an effect on saw palmetto density. The single application of a dormant or growing season burn is not recommended for control of high-density saw palmetto, however, it may be suitable to maintain areas where saw palmetto levels are low and proliferation of the species is not a threat. Dormant and growing season roller chopping showed the greatest potential for rapid saw palmetto control. Growing season roller chopping is recommended if significant reductions in saw palmetto density are desired.     

Regeneration dynamics in remnant Tsuga canadensis stands in the northern Lake States: Potential direct and indirect effects of herbivory

We examined the effects of white-tailed deer (Odocoileus virginianus) herbivory and microsite limitation on Tsuga canadensis regeneration in 39 randomly selected remnant T. canadensis stands in Michigan's Upper Peninsula. Deer of the region migrate to and congregate in T. canadensis stands in winter resulting in strong seasonal habitat use patterns. In each study stand, we quantified vegetation, microsite availability, and deer use (via pellet counts). While some stands contained high densities of T. canadensis regeneration (stems < 4.0 cm dbh), we found complete T. canadensis regeneration failures in 6 out of 39 stands. Additionally, 17 and 22 stands respectively, had complete failures in the small and large sapling categories. General linear models suggested that deer use was the primary limiting factor in the small sapling size class, even at relatively low levels of deer use. T. canadensis seedling density was positively associated with the availability of high-decay coarse woody debris and negatively associated with basal area of Acer saccharum in the overstory. This latter association may be due, at least in part, to negative effects of broadleaf litter on T. canadensis establishment and a general trend toward increasing Acer abundance in the regeneration layer. Our results suggest that differential tolerance to browsing (Tsuga vs. Acer) in conjunction with reduced germination substrate availability may set up a scenario where successful T. canadensis establishments is more limited by legacy and indirect than direct effects given contemporary levels of deer use.     

The effects of thinning and burning treatments on within-canopy variation of leaf traits in hardwood forests of southern Ohio

Leaf nitrogen content (N_mass, %) and leaf mass per area (LMA, g m⁻²) are two important features that are closely linked to the photosynthetic performance of plants and, thus, the NPP of forest ecosystems. Forest management practices, such as burning and thinning, change stand structure and soil dynamics, which may result in changes in N_mass and LMA. The objective of this study was to understand how N_mass and LMA of seven canopy tree species/genus (Quercus alba, Q. coccinea, Q. prinus, Q. velutina, Carya spp., Acer rubrum, and Liriodendron tulipifera) responded to (i) thinning and/or burning treatments and to (ii) different landscape and soil properties in southern Ohio. We collected leaves from the top, and bottom, of five individuals of each taxa in each treatment unit. Leave traits (N_mass and LMA) were compared using analysis of variance followed by orthogonal contrasts. To further understand the factors that influence the canopy leaf traits, we used regression tree analysis (RTA) to examine how variations of LMA and N_mass were linked to thinning and/or burning treatments, soil, and landscape variables. Finally, we assessed the potential ramifications of changes in these traits on canopy carbon budgets using a PnET-Day model, which is a daily time-step canopy carbon exchange model. We found significant effects of thinning, burning, and their interactions on LMA at the bottom of the crown while none of the treatments showed significant effects on LMA at the top of the crown. N_mass responded significantly to only burning treatment. RTA results exhibited minor effects of landscape features and soil properties on N_mass and LMA. Interspecific differences accounted for most variations of both leaf traits. Sensitivity analysis of PnET-Day model suggested these subcanopy changes in LMA increased the annual net primary production (NPP) by 8%. In summary, our results suggest that forest management can substantially influence canopy leaf traits such as N_mass and LMA and that alteration of these traits can influence forest NPP. Given the role of forests as global carbon sinks, the potential influence of thinning and burning on canopy traits, and thus NPP, is an important consideration for forest management.     

An individual-tree diameter growth model for managed uneven-aged oak-shortleaf pine stands in the Ozark Highlands of Missouri, USA

The Pioneer Forest encompasses more than 60,000 ha in the Ozark Highlands of Missouri, USA and has been managed using single-tree selection since the early 1950s. This paper quantifies the influence of tree size and competitive position, stand density, species composition, and site quality on ten-year (1992-2002) diameter increment within oak (Quercus spp.) and shortleaf pine (Pinus echinata Mill.) stands on the Pioneer Forest. An individual-tree model was developed for each species using mixed-effects regression and 290 inventory plots. Model efficiency (R²) ranged from 0.26 to 0.57 and fit was generally better for oak species. Basal area in larger trees (BAL) and tree diameter were significant predictors for all species and crown competition factor improved prediction for shortleaf pine and hickory (Carya spp.). Effect of species composition and site quality on diameter growth was not consistent across species. Models were evaluated using a subset of data not included in model fitting and the effect of single tree and standwise (1, 3, or 5 sample trees) calibration on model predictions were evaluated. Inclusion of random effects through calibration improved model prediction for all species and fit was best following single tree and 3 tree calibration.     

Short term effects of alternative silvicultural treatments on stand attributes in irregular balsam fir-black spruce stands

The eastern Canadian boreal forest exhibits a specific disturbance regime where forest fires are less frequent than in the western part. This particularity may explain the abundance of irregular stands with distinct ecological features. To ensure sustainable forest management, these characteristics require the implementation of an adapted silviculture regime. In this context, two selection cutting methods were developed and compared with more conventional techniques, initially designed for cutting more regular stands of the boreal forest (cutting leaving small merchantable stems, careful logging preserving advance regeneration). The comparison focused on the capacity of treatments to maintain the primary attributes of irregular boreal forests, including complex vertical structure, abundant tree cover, species composition, and an abundance of dead wood. Mortality and regeneration processes were also compared.     

Effects of bamboo stands on seed rain and seed limitation in a rainforest

Bamboos often negatively affect tree recruitment, survival, and growth, leading to arrested tree regeneration in forested habitats. Studies so far have focused on the effects of bamboos on the performance of seedlings and saplings, but the influence of bamboos on forest dynamics may start very early in the forest regeneration process by altering seed rain patterns. We tested the prediction that the density and composition of the seed rain are altered and seed limitation is higher in stands of Guadua tagoara (B or bamboo stands), a large-sized woody bamboo native from the Brazilian Atlantic Forest, compared to forest patches without bamboos (NB or non-bamboo stands). Forty 1 m² seed traps were set in B and NB stands, and the seed rain was monitored monthly for 1 year. The seed rain was not greatly altered by the presence of bamboos: rarefied seed species richness was higher for B stands, patterns of dominance and density of seeds were similar between stands, and differences in overall composition were slight. Seed limitation, however, was greater at B stands, likely as a resulted of reduced tree density. Despite such reduced density, the presence of trees growing amidst and over the bamboos seems to play a key role in keeping the seeds falling in B stands because they serve as food sources for frugivores or simply as perches for them. The loss of such trees may lead to enhanced seed limitation, contributing ultimately to the self-perpetuating bamboo disturbance cycle.     

Modelling jay (Garrulus glandarius) abundance and distribution for oak regeneration assessment in Mediterranean landscapes

Because natural regeneration of oak is strongly dependent on jay abundance and distribution, we need to understand the determinants of jay abundance and occurrence as a first step to assess oak regeneration. In this paper we modelled the jay population distribution and abundance in a Mediterranean landscape mosaic of the eastern Iberian Peninsula (Valencia, east Spain). The methodology was based on assessing landscape attributes (habitat composition and configuration variables) on seven 9-km² sites and registering jay fledgling locations. Using a stepwise regression model we determined the variables that best explained the jay density. The probability of occurrence of jay nests within each site was assessed by comparing the landscape attributes of jay and non-jay areas within each site. Results were validated by predicting each site on the basis of data from the remaining six sites and then calculating the deviation between the predicted and the observed values in the field. The results suggest that jay density correlates positively with forest cover and landscape heterogeneity variables, and negatively with shrubland cover. Validation of the results showed that the model is reasonably effective in predicting both jay abundance and distribution at the extent and resolution used.     

Effects of prescribed fire and season of burn on direct and indirect levels of tree mortality in Ponderosa and Jeffrey Pine Forests in California,USA

Many forests that historically experienced frequent low-intensity wildfires have undergone extensive alterations during the past century. Prescribed fire is now commonly used to restore these fire-adapted forest ecosystems. In this study, we examined the influence of prescribed burn season on levels of tree mortality attributed to prescribed fire effects (direct mortality) and bark beetles (Coleoptera: Curculionidae, Scolytinae) (indirect mortality) in ponderosa pine, Pinusponderosa Dougl. ex Laws., and Jeffrey pine, Pinusjeffreyi Grev. and Balf., forests in California, USA. A total of 816 trees (9.9% of all trees) died during this 3-yr study. Significantly higher levels of tree mortality (all sources) occurred following early and late season burns compared to the untreated control, but no significant difference was observed between burn treatments. The majority (461 trees) of tree deaths were attributed to direct mortality from prescribed burns and was strongly concentrated (391 trees) in the smallest diameter class (<20.2 cm diameter at breast height, dbh). For the largest trees (>50.7 cm dbh), significantly higher levels of tree mortality occurred on early season burns than the untreated control, most of which resulted from indirect mortality attributed to bark beetle attacks, specifically western pine beetle, Dendroctonus brevicomis LeConte, and mountain pine beetle, D. ponderosae Hopkins. Red turpentine beetle, D. valens LeConte, was the most common bark beetle species found colonizing trees, but tree mortality was not attributed to this species. A total of 355 trees (4.3% of all trees) were killed by bark beetles. Dendroctonus brevicomis (67 trees, 18.9%) and D. ponderosae (56 trees, 15.8%), were found colonizing P. ponderosa; and Jeffrey pine beetle, D. jeffreyi Hopkins, was found colonizing P. jeffreyi (seven trees, 2.0%). We also found pine engraver, Ips pini (Say) (137 trees, 38.6%), and, to a much lesser extent, Orthotomicus (=Ips) latidens (LeConte) (85 trees, 23.9%) and emarginate ips, I. emarginatus (LeConte) (3 trees, 0.8%) colonizing P. ponderosa and P. jeffreyi. Few meaningful differences in levels of indirect tree mortality attributed to bark beetle attack were observed between early and late season burns. The incidence of root and root collar pathogens (Leptographium and Sporothrix spp.), including species known to be vectored by bark beetles, was low (18% of trees sampled). The implications of these and other results to management of P. ponderosa and P. jeffreyi forests are discussed in detail.     

The effects of burning and dead-wood creation on the diversity of pioneer wood-inhabiting fungi in managed boreal spruce forests

The loss of natural forest habitats due to forestry is the main reason for the decline in boreal forest biodiversity of the Nordic countries of Europe. Ecological rehabilitation may provide means to recover and sustain biodiversity. We analyzed the effects of controlled burning and dead-wood creation (DWC) on the diversity of pioneer wood-inhabiting fungi in managed Norway spruce (Picea abies) forests in southern Finland. Altogether 18 stands were first subjected to a partial cutting with ordinary logging residues in form of cut stumps and treetops left on site. The subsequent rehabilitation treatments consisted of a controlled burning applied in half of the stands and three levels of dead-wood creation (5, 30 and 60 m³ ha⁻¹). The DWC involved creation of logs; felling of whole trees to mimic downed logs formed by natural disturbance processes. Each treatment was replicated three times. Inside each stand, substrates were sampled in two different biotopes; one on mineral soil and one on mineral soil with a thin peat layer. We surveyed the fungal flora on the logs (n = 364) and the ordinary residue stumps (n = 1767) and tops (n = 845) five years after the treatments.When comparing different stands, controlled burning had a significant effect on species composition; certain species were significantly more frequent on substrates in burned stands than in unburned stands, indicating that these species were favored by controlled burning. By contrast, we found no significant effects of DWC levels or biotope on species composition or richness. When comparing different substrates, 99% of the logs hosted at least one species and the occurrence probability of certain species was significantly higher on logs than on ordinary residue stumps and tops. Yet, volume-based rarefaction analyses showed that residues were more species dense than the logs, indicating that ordinary logging residues constitute important resources for many pioneer species.We conclude that controlled burning combined with DWC have strong effects on biodiversity; it modifies the composition of the pioneer wood-inhabiting fungal species found in managed forests and may thereby also influence the further succession and diversity of the secondary fungal flora.     

Short-term effects of fire on soil properties inPinus densiflora stands

This study was performed to investigate a short time change (one week after fire) on soil properties due to the fire inPinus densiflora Sieb. et Zucc stands of the Kosung area in Kangwon Province in Korea. Twenty seven sampling plots [16 burned (8 low intensity fire, 8 high intensity fire) and 11 unburned plots] were chosen. Mineral soil samples from three depths (0–5, 5–15, and 15–25 cm) under the forest floor were collected. Forest fire in the area affected soil chemical properties. Soil pH, total nitrogen, available phosphorus, potassium, calcium, and magnesium in the surface soil (0–5 cm) of the burned area compared with the unburned area increased, but there was no marked change in the subsurface soil (5–25 cm). Organic matter, total nitrogen, available phosphorus, and exchangeable cations in the surface soil were generally lower in the high than in the low intensity fire areas. This indicates that these nutrients on the high intensity fire may be volatilized. The results suggest that change in soil chemical properties in the area was restricted mainly to the surface soil and was different between the high and the low intensity fire types.     

Differences in ecophysiological responses to summer drought between seedlings of three deciduous oak species

Seasonal variations in water status and gas exchange were assessed in 2-year-old seedlings of Quercus pubescens Willd., Quercus ithaburensis Decaisne subsp. macrolepis (Kotsch) Hedge & Yalt, and Quercus frainetto Ten. during the 2004 growing season characterized by the typical summer drought conditions observed in Greece. The seedlings were grown under field conditions. After establishment half of the seedlings were well watered, while the rest received no irrigation other than natural rainfall. Measurements were conducted from April to October.     

Canopy gaps and regeneration in old-growth Oriental beech (Fagus orientalis Lipsky) stands, northern Iran

Virgin beech Fagus orientalis forests in northern Iran provide a unique opportunity to study the disturbance regimes of forest ecosystems without human influence. The aim of this research was to describe characteristics of natural canopy gaps and gap area fraction as an environmental influence on the success of beech seedling establishment in mature beech stands. All canopy gaps and related forest parameters were measured within three 25 ha areas within the Gorazbon compartment of the University of Tehran’s Kheyrud Experimental Forest. An average of 3 gaps/ha occurred in the forest and gap sizes ranged from 19 to 1250 m² in size. The most frequent (58%) canopy gaps were <200 m². In total, canopy gaps covered 9.3% of the forest area. Gaps <400 m² in size were irregular in shape, but larger gaps did not differ significantly in shape from a circle. Most gaps (41%) were formed by a single tree-fall event and beech made up 63% of gap makers and 93% of gap fillers. Frequency and diversity of tree seedlings were not significantly correlated with gap size. The minimum gap size that contained at least one beech gap-filling sapling (<1.3 m tall) was 23.7 m². The median gap size containing at least one beech gap-filling sapling was 206 m² and the maximum size was 1808 m². The management implications from our study suggest that the creation of small and medium sized gaps in mixed beech forest should mimic natural disturbance regimes and provide suitable conditions for successful beech regeneration.