Disturbances are more likely to occur in previously damaged stands (Hanewinkel et al., 2011; Ruel et al., 2023). Surviving trees have often encountered mechanical damage to roots, xylem, and crown, stressing them and increasing their susceptibility (White and Pickett, 1985; Puhe, 2003; Seidl and Blennow, 2012; Harmon and Pabst, 2019). Stress from injuries and abrupt changes in environmental conditions might cause growth retardation. Seidl and Blennow (2012) found that the increment of Norway spruce (Picea abies L. Karst) dropped significantly in forests affected by the highly destructive winter storm Gudrun, compared to pre-storm levels. Exposed remnant trees experience physiological adaptations including thickening of trunks and increased root growth; response rate depends on growth conditions (Stathers et al., 1994). Trees may need several years to acclimate to an altered wind environment and regain mechanical stability. Mortality from other agents often remains high during these years (Schroeder and Lindelöw, 2002; Köster et al., 2009; Shibuya and Ishibashi, 2019) as trees allocate resources to repair mechanical damages at the expense of defensive mechanisms, thereby increasing their susceptibility to insect pests and pathogens (Schowalter, 1985; Baier et al., 2002; Wermelinger, 2004; Huang et al., 2020; Puglielli et al., 2023).

Delayed mortality can strike trees without any visible damage. Ueda and Shibata (2004) associated tree decline and dieback of apparently healthy-looking remnant trees in a typhoon-damaged forest stand with increased aboveground hydraulic resistance caused by a large reduction in trunks’ water conducting area. Although delayed mortality mostly occurs within a few years following the disturbance, Harmon and Pabst (2019) observed delayed mortality long after the damaging wind event—of 319 surviving wind-damaged trees, 144 (about 45%) died within 35 years. Mortality of the damaged trees was highest within the first 5 years, but approximately 11% of mortality cases occurred after a 10–25-year delay.

Wind-damaged trees can sometimes resprout, rebuild injured crowns or develop secondary shoots from downed stems, even despite severe damage (Peterken, 1996; Cooper-Ellis et al., 1999; Peterson, 2000; Canham et al., 2001). Sprouting ability is substantially affected by several tree traits (e.g., species, age, size) as well as by the level of pre-disturbance competition between trees (Peterson and Carson, 1996; Peterson and Rebertus, 1997; Cooper-Ellis et al., 1999; Peterson, 2000; Dietze and Clark, 2008; Matula et al., 2019). Type of damage (stem breakage versus uprooting) is also important—uprooted trees sprout more often than broken trees (Dyer and Baird, 1997). As most conifers do not sprout, stand regeneration via vegetative recovery is usually unattainable in wind-damaged conifer-dominated stands, while in situations where species prone to sprouting make up a large proportion of the forest, the role of sprouting in re-establishment of the forest structure could be substantial (Peterson, 2000; Dietze and Clark, 2008).

Peterson and Pickett (1991) reported that 25% of broken trees sprouted in the four growing seasons after a tornado in an old-growth hemlock-hardwood forest; the tendency to sprout varied significantly among species. Of those broken trees that sprouted, 68% were alive at the end of the fourth growing season. Cooper-Ellis et al. (1999) reported that a surprisingly large amount (approximately 80%) of damaged trees survived and sprouted in the first growing season in an experimentally pulled down 75-yr-old temperate Quercus rubra–Acer rubrum forest. After 4 years, survival of uprooted and broken trees remained above 40% (Foster et al., 1997). In a follow-up study, Plotkin et al. (2013) concluded that the presence of abundant advance regeneration, root suckers and stump sprouts, understory vegetation, and, most importantly, surviving trees allowed the forest to resist changes in ecosystem processes and functions. However, when a forest stand is mainly recovered via sprouting, an abrupt shift in tree-community structure and composition can occur, due to the differential ability of species to sprout (Peterson and Pickett, 1991; Everham and Brokaw, 1996). Moreover, sprouts (and trees that originate from sprouts), may be more susceptible to wood-rotting organisms and further mechanical damage, affecting the future development of the stand (True and Tryon, 1966; Putz and Sharitz, 1991; Del Tredici, 2001).

Characteristics of a wind event also affect the condition of remnant trees. A simulation of wind damage to old-growth boreal stands of mixed species showed that wind speed was more important than the number of events for relatively low wind speeds (equal to or < 20 m s⁻¹), but at higher wind speeds, the number of events had greater impact on severity (i.e., damage done to the forest in terms of proportion of trees downed, loss in basal area, loss in above-ground biomass; Anyomi et al., 2016). For long duration events, damage propagates until all susceptible trees in a stand are windthrown. Similarly, frequent high wind speed events generally clear an area more rigorously of weaker trees than do low wind speed events, regardless of their frequency. Short duration high wind speed events, however, create patches of damage that are bordered by stand edges that survived only because the wind speed dropped. These edges remain vulnerable to subsequent winds of similar magnitude (Anyomi et al., 2016).

Downbursts are a special case of wind disturbance that can result in distinct patterns of wind damage. A downburst is a rapid, localized, straight-direction surface wind caused by a strong downdraft from the base of convective thundershowers and thunderstorms (Xi and Peet, 2011). Contrary to other types of wind disturbances, in case of a downburst, the trees sheltered by other trees or topographic features may not have advantages over the trees that have more exposed positions, as all trees in a path of a downburst regardless of their level of exposure are similarly vulnerable to the downward facing wind pressure (Foster et al., 1998; Lin et al., 2004). Thus, downburst often create sharp edges between intact forest and completely windthrown areas (Foster et al., 1998).

Wind conditions within the first years after the initial event are critical in determining the long-term survival of remnant trees as air flow at higher speeds in more exposed stands may cause further damage to crowns, stems, or root systems of remnants (Hunter, 1995; Mitchell, 2013). For instance, Kitenberga et al. (2021) found that birch trees with prior damage had a significantly higher probability (odds ratio 4.32, i.e., four times as likely) of further wind damage. Vulnerability of the remnant stand to further wind damage generally increases with increasing severity of the initial event, whereas the level of damage from following wind event can be much higher than expected for the level of winds experienced (Gardiner, 2021).

The timing of the next wind event is important. If the wind event takes place between growing seasons, conifers are more susceptible to windthrow than deciduous broadleaves due to needle retention (Zeng et al., 2010). Moreover, root anchorage strength is increased by soil freezing and reduced by soil saturation (Gardiner et al., 2010), thus trees are more susceptible to uprooting if wind disturbance takes place when the ground is saturated but not yet frozen. Precipitation during a windstorm also increases susceptibility to wind damage as water adds extra weight to the canopy, increasing the bending forces onto the stem when the stem is displaced by the wind (Gardiner et al., 2008).

Long-term disturbance and land use history affect the susceptibility of the survivors to further wind damage. For instance, even though susceptibility to wind damage generally increases with increasing age and size of trees (Foster, 1988; Whitney, 1989), sometimes the surviving trees may be the oldest and largest ones in the stand (Everham and Brokaw, 1996; Peterson, 2000). Very large trees can be open-grown trees from previous land use or trees that have had time to develop open-grown characteristics when they became canopy emergents due to the gradual death of surrounding trees. Such combinations of legacies that often contribute to increased resilience of the forest are termed ‘legacy syndromes’ (Jõgiste et al., 2017). On a single tree scale, heavy snow or ice loading on the crown, which occurs primarily in forests located at higher latitudes or high altitudes, increases windthrow susceptibility due to increased canopy mass and an increase in the drag coefficient (Valinger et al., 1993; Stathers et al., 1994). Concomitant factors, such as falling of adjacent trees or branches can also cause bark, root, crown, and sapwood damage, in some cases leading to fatality (Webb, 1989; Cooper-Ellis et al., 1999).

The increased amounts of CWD associated with wind disturbances increase the probability and severity of a subsequent wildfire; wind–fire interactions vary with climatic conditions, with regional differences, and with the intensity or severity of individual disturbances (Kulakowski and Veblen, 2007; Cannon et al., 2017). Wind disturbance may interact with fire not only through the addition of fuel, but also through more subtle changes in fuel composition, consumption, and arrangement (Cannon et al., 2014). Moreover, wind damage may increase fire intensity or severity indirectly as well, by changing species composition and/or forest structure toward more flammable ones (Cannon et al., 2017; Anoszko et al., 2022).

Insect pests are the most significant biotic mortality-causing agents following a wind disturbance. Increased amounts of CWD in wind-disturbed forests are prime breeding habitats for potentially greatly damaging subcortical insects (Gandhi et al., 2007). In turn, insect pests can vector fungal infections that aid in the depletion of host defenses (Paine et al., 1997; Krokene and Solheim, 1998; Faccoli and Santini, 2016; Vogt et al., 2020). Bacteria also may play a role in neutralizing tree defenses (Boone et al., 2013).

Tree-killing bark beetles can have a profound ecological effect on wind-disturbed remnant stands, influencing their species composition, age structure, density, woody debris inputs etc. (Raffa et al., 2015). At stand scale, local context regarding insect pests (e.g., presence and abundance of the pest, climatic conditions influencing reproduction rate of pests as well as susceptibility of a host) is clearly important; presence and abundance of antagonists, i.e., predators and parasites of the pest, may also make a considerable difference (Massey and Wygant, 1954; Osetrov, 2002; Bouget and Duelli, 2004; Wermelinger, 2004; Eriksson et al., 2007; Kärvemo et al., 2014a; Jan Weslien et al., 2024). Location of the disturbed forest can influence the severity of infestation as voltinism (number of broods or generations of an organism within a year) of insect pests depends on latitude and elevation, with a greater number of generations occurring in milder climates (Lange et al., 2006; Jönsson A M, et al., 2007). Species composition of the disturbed stand is of high importance, as pests and pathogens are usually host-specific (Bouget and Duelli, 2004; Tedersoo et al., 2019). There is a strong positive correlation between stand-scale host tree volume and risk of bark beetle infestation (Kärvemo et al., 2014b). Among bark beetles that colonize live trees, most have evolved adaptations to exploit conifers in the Pinaceae family, despite remarkable defenses that these trees can mount (Sequeira et al., 2000; Franceschi et al., 2005; Raffa et al., 2015). Species composition can have more complex effects on vulnerability as well, for example, Baier et al. (2002) found that spruces in mixed stands had a higher primary resin flow as a response to bark beetle attack, compared to trees in pure spruce stands. Moreover, it has been found that volatiles released by non-host trees disturb bark beetle response to pheromones (Byers et al., 1998; Jactel et al., 2001; Zhang, 2003), thus possibly improving the resistance of mixed forests.

On a single tree scale, the level of damage the tree was subjected to during wind disturbance determines the fate of a remnant tree to a substantial extent, as bark beetles can detect cues associated with stress physiology of their host plants (Raffa et al., 2015). Attacks by several insect species can occur simultaneously, causing cumulative effect (Göthlin et al., 2000). Higher entry rates have been observed in response to root infection, defoliation, fire- or lightning injury, old age, and other stresses (Christiansen, 1991; Jakuš et al., 2011; Lindgren and Raffa, 2013). However, once the pest population has reached the epidemic phase, host defenses no longer play a major role since the insects are abundant enough to overcome the defense mechanism in all but the most vigorous or genetically resistant trees (Lindgren and Raffa, 2013). The age of a tree is a significant factor, as several age-related parameters (thickness and texture of bark, thickness of phloem, vitality of the host) affect its susceptibility to bark beetles (Långström, 1983; Ferrenberg and Mitton, 2014; Takei et al., 2021), whereas different species have dissimilar host requirements (Göthlin et al., 2000). For instance, the lower limit of the thickness of bark which can be colonized by Ips typographus L. is 2.5 mm, which is usually attained when the DBH of the host tree has reached about 15 cm (Grunwald, 1986; Seidl and Rammer, 2017). Dendroctonus rufipennis Kirby prefers large, mature trees as a host, but may attack trees with DBH as small as 5 cm when larger trees become depleted and the beetle population has become excessively abundant (Massey and Wygant, 1954). Some highly damaging bark beetles prefer a breeding substrate exposed to the sun; Jakuš et al. (2011) found that trees with a higher level of stem shading (longer crown length) had a higher probability of surviving the I. typographus outbreak. Some observations of potentially highly damaging insect pests associated with wind disturbances in temperate and boreal biomes are described in Table 1. Several landscape and stand-scale parameters (e.g., elevation, aspect, initial stand density) additionally affect the predisposition of the remnant stand to insect outbreaks (Tables 2, 3).

The decline-disease spiral model concept applies in regard to insect pest outbreaks, as all large-scale wind disturbances within the range of severe pests such as D. rufipennis or I. typographus do not result in subsequent outbreaks, indicating that some other factors need to co-occur for a large-scale infestation to emerge; drought has been suggested to be the most influential contributing factor (Jönsson A M, et al., 2007; Raffa et al., 2008; Hart et al., 2014; Netherer et al., 2015; Dodds et al., 2019). For instance, after World War II, catastrophic insect outbreaks occurred when large cuttings for reparations that generated huge amounts of litter and debris were followed by a hot and dry summer in 1946 (Hanewinkel et al., 2011).

The presence of rot-causing fungi such as Heterobasidion spp., Armillaria spp. and Phellinus spp. is known to harm the mechanical stability of infected trees, thus increasing the risk of wind damage (Whitney, 1989; Quine and Gardiner, 2007). In turn, wind disturbance may facilitate the spread of pathogenic fungi by creating infection courts on residual trees from collision and abrasion from falling trees, by fracturing the roots, and by fungal spore dissemination (Schmid-Haas and Bachofen, 1991; Hennon, 1995; Worrall et al., 2005; Krisans et al., 2020). Moreover, wind disturbances can facilitate fungal pathogens via altered microclimatic conditions (e.g., temperature, relative humidity) which can increase abundance and infection success of many pathogenic species (Cobb and Metz, 2017; Idbella et al., 2023). Additionally, wind disturbance can substantially alter the communities of symbiotic fungi (Idbella et al., 2023; Venice et al., 2023), which can affect seedling establishment (Nara, 2006; Liang et al., 2020) and decrease stand’s resistance to abiotic and biotic stressors (Van Der Heijden et al., 2008; Anthony et al., 2022). However, the impact of wind disturbance on fungal communities is not a subject of many research papers, thus the magnitude and causality of these interactions is still largely unclear.

Due to different susceptibility of tree species, a forest stand with diverse tree species composition is less vulnerable to fungal pathogens, compared to monospecific stands (Pautasso et al., 2005). Some studies have shown that root and butt rot (caused by Heterobasidion spp. and other pathogens) frequency in Norway spruce is higher in fertile sites (e.g., Garbelotto and Gonthier, 2013; Müller et al., 2018).

Bark beetles that colonize living conifers are frequently associated with pathogenic fungi that are introduced into the tree during the attack process (Paine et al., 1997; Krokene and Solheim, 1998; Netherer et al., 2021). These symbiotic fungi are important allies to the pest as they produce beetle semiochemicals and degrade host toxins, helping to exhaust tree defenses (Netherer et al., 2021). Among the fungal associates of spruce bark beetles in Eurasia, a blue stain fungus Endoconidiophora polonica (formerly known as Ceratocystis polonica), primarily transmitted by I. typographus, is the most virulent (Solheim, 1988; Christiansen, 1991; Krokene and Solheim, 1998.). Another blue stain fungus Grosmannia penicillata is also a noteworthy subsidiary to the pest (Schebeck et al., 2023; Netherer et al., 2024). In North America, Leptographium abietinum is associated with Dendroctonus spp., causing additive damage to members of the Pinaceae, especially Picea spp. (Jacobs et al., 1998). Tomicus spp. have several fungal associates as well, for instance they are an active transmitter of the tracheomycotic genus Ophiostoma, inducing blue stain in Pinus yunnanensis Franchet in China (Pan et al., 2017) and Pinus sylvestris in Sweden (Solheim and Långström, 1991). In northern Spain, Tomicus piniperda has been found to vector Fusarium circinatum, the causal agent of pitch canker disease, in Pinus radiata D. Don (Bezos et al., 2015).

Disturbance size and severity are inversely related to event frequency (Foster and Reiners, 1986; Turner et al., 1998; Angelstam and Kuuluvainen, 2004; Hanewinkel et al., 2011). This relationship creates a gradient from infrequent, coarse-grained stand-replacing disturbances (SR) to moderately frequent, medium-grained disturbances (partially stand-replacing disturbances, PR), to very frequent fine-grained single (or few) tree gap disturbances (fine-scale gap disturbances, FS), as depicted in Figure 1. These regimes usually create even-aged stands, multi-aged stands, and all-aged stands, respectively (Frelich, 2002), according to the prevailing local disturbance regime. Within the framework of this study, we consider a wind disturbance event to be stand-replacing when >75% of the upper strata (i.e., canopy trees and emergent trees) are downed during the incident, as proposed by Bouchard et al. (2009). Fine-scale gap disturbance occurs when small gaps are formed by one or a few fallen trees, this disturbance regime is inherent in late-successional forests (Runkle, 1985; Kuuluvainen, 1994). Gap size for single treefalls mostly ranges from 50 to 200 m² (McCarthy, 2001). Discriminating FS from PR is challenging, as the threshold between the severity levels is often unclear (Hart and Kleinman, 2018). Within the framework of this review, we consider a wind disturbance to be partially stand-replacing when at least 15% (but less than 75%) of the trees in the upper and mid canopy layers have been downed during the event. Our delineation is based on a review article (McCarthy, 2001) in which gap characteristics of gap-disturbed forests were examined. We acknowledge that setting such a threshold is arbitrary, thus, if possible, we also took origin of the gaps into consideration: partially stand replacing disturbances are abrupt exogeneous events, while fine-scale gap disturbances are more gradual and mostly driven by endogenous agents such as fungi, insects, and inter-tree competition; wind appears as the final agent that fells the predisposed trees.

Long-term estimates regarding the fate of remnant trees are rarely documented, but Shibuya and Ishibashi (2019) reported a cumulative 56% mortality rate of remnant trees 60 years after a wind event that downed 93% of the basal area (85% of trees) in a conifer-dominated study plot located in Hokkaido. In a mixed species study plot nearby where the same wind event had downed 86% of the basal area (82% of trees), the cumulative post-windthrow mortality added up to 60% 37 years after the wind disturbance (mortality was not tracked from there on). In both study plots, mortality was particularly high during the first 5 years following the windthrow event, especially for conifers. It must be kept in mind that after an extended period of time, tree death cannot always be clearly associated with wind disturbance anymore, as with an expanding timescale, other interacting factors (e.g., other unrelated abiotic and biotic disturbance agents, tree senescence) become increasingly influential (Harmon and Bell, 2020).

The mortality probability of a remnant stand depends on the severity of the disturbance; remnants from stand-replacing disturbances are generally more susceptible to subsequent disturbances than remnants from less severe disturbances. Evidence regarding high correlation between disturbance severity and growth retardation of remnant trees (Seidl and Blennow, 2012) indicates that trees endure more stress after high-severity wind disturbance, compared to a less severe event. The authors found that for every 10% of the standing timber volume structurally damaged by storm Gudrun, a 6.7% growth reduction occurred on average in the 3 years following the storm.

Severe large disturbances create more variability in successional pathways than small disturbances, providing opportunities for the initiation of multiple stable states, while abundance and spatial arrangement of remnants is a key factor in determining the outcome (Turner et al., 1998). Wind disturbances are rarely spatially uniform; thus, the heterogeneity of storm impacts can cause the spatial pattern of survivors to be more aggregated compared to a pre-disturbance state, whereas aggregation is likely to peak at high but not extreme levels of severity (Woods, 2004; Peterson, 2020). For instance, Wolf et al. (2004) observed strongly aggregated mortality patterns at scales of 5–50 m as a result of the 1999 extratropical cyclone in a Danish mixed-species deciduous forest. Allen et al. (2012) also reported that a severe storm greatly altered the spatial pattern of a Pinus strobus stand, producing a more clumped distribution of surviving trees. Such development was not clear in an oak-dominated (Quercus borealis) stand nearby that was similar in topographic and soil conditions but had endured less severe wind damage. The tendency for the survivors to form aggregated patterns can be important, because numerous studies on variable retention systems have demonstrated that in general, the proportion of residual trees left and their onward wind-caused mortality are inversely related, while highly dispersed trees have higher mortality than the trees clustered in aggregates (e.g., Moore et al., 2003; Maguire et al., 2006; Thorpe and Thomas, 2007; Lavoie et al., 2012; Urgenson et al., 2013). Similarly, Peterson and Cannon (2021) found that support from neighboring trees reduces the force that a given tree must withstand, thus increasing critical wind speed required to down the tree.

Survivors growing at the formerly open edges of a wind-damaged stand are better acclimated and generally less susceptible to the next wind, while newly exposed trees that previously grew in the stand interior are not accustomed to the exposure and thus more susceptible (Ennos, 1997; Brüchert and Gardiner, 2006; Gardiner et al., 2008). Static wind load on survivors is positively correlated with the size of a canopy opening; the increase of maximum static wind load is up to 14 times higher than the load in an undisturbed forest, thus possibly leading to onward windthrow mortality at the downwind gap edges (Quine and Gardiner, 2007; Panferov and Sogachev, 2008). Moreover, the maximum static wind load is experienced at the downwind gap edge independently of gap size, the maximum turbulent kinetic energy, however, shows the apparent shift toward the lateral and even to the leeward gap edges with the increasing gap diameter above a size of 15 times canopy height (Panferov and Sogachev, 2008), making all remnant trees at newly formed edges of a large gap susceptible to next heavy intense wind event. On the other hand, recurring windstorms may have only a little effect on a forest if the previous storm has already removed most of the susceptible trees (Webb, 1989).

High-severity wind disturbances are accompanied by a greater amount of breeding material with non-existent or weak defenses and more favorable microclimatic conditions for bark beetles, compared to less severe disturbances (Göthlin et al., 2000; Schroeder and Lindelöw, 2002; Wermelinger, 2002; Kärvemo et al., 2014b). Some hazardous forest pests such as I. typographus attack windthrown spruces in gaps more frequently than they attack trees along edges, which in turn are preferred over trees in closed stands (Göthlin et al., 2000; Schroeder and Lindelöw, 2002; Bouget and Duelli, 2004). Mortality caused by I. typographus usually peaks in the second or third summer after severe wind damage—insects migrate from gap interior to the forest edge where they first colonize the most susceptible trees, but later, when the bark beetle population size has reached epidemic level, they may migrate to trees without any visible damage (Schroeder, 2001; Osetrov, 2002; Schroeder and Lindelöw, 2002; Kärvemo et al., 2014a). For instance, Köster et al. (2009) found that only 25% of the initial Norway spruce remnants that survived at the edges of the largest wind gaps were alive 5 years after the wind event, whereas I. typographus caused most of the subsequent tree deaths. South-facing newly formed gap edges have been found to be twice as vulnerable to I. typographus attacks, compared to north-facing gap edges (Kautz et al., 2013).

The magnitude of bark beetle damage can be immense—in western Colorado, after a severe windstorm of 1939, a severe outbreak of D. rufipennis followed, destroying an additional 10.1 million m³ of timber by 1951 (Massey and Wygant, 1954). The outbreaks of D. pseudotsugae can be similarly destructive—after a series of windstorms between 1949 and 1953 in western Oregon and southwestern Washington that resulted in 27 million m³ of windthrown trees, an additional 8.3 million m³ of standing trees were killed by D. pseudotsugae (Gandhi et al., 2007).

Depending on spatial and temporal factors, wind-disturbed forests may be highly vulnerable to subsequent fire disturbance (Tables 2–4); vulnerability increases with increasing severity of wind damage (Cannon et al., 2017). For instance, due to the increase in CWD, modeled fire burn times and the extent of the fire were found to increase dramatically with increasing blowdown severity (Buma and Wessman, 2011). Similarly, Kulakowski and Veblen (2007) found that high blowdown severity was strongly spatially correlated with high fire severity. The risk of fire substantially increases if wind disturbance is followed by periods of dry, hot, windy weather (Kulakowski and Veblen, 2007; Mitchell, 2013).

Large areas of boreal and temperate forests are naturally regulated by moderate-severity disturbances, leading to heterogeneous and dynamic stand and landscape structures (Frelich and Lorimer, 1991; Papaik and Canham, 2006; Kuuluvainen, 2009; Stueve et al., 2011; Koivula et al., 2014; Khakimulina et al., 2016; Hart and Kleinman, 2018; Meigs and Keeton, 2018). Generally, structural complexity peaks when wind disturbance is of intermediate severity (Peterson, 2019b). Greater stand-scale structural and compositional diversity enables spreading risk among survivors that have different susceptibility to various stressors and disturbance agents (Jactel et al., 2017; Mitchell et al., 2023). Thus, if such heterogeneity is evident following a moderate-severity wind disturbance, the survivors are generally more resistant to subsequent disturbance agents such as wind, insects, pathogenic fungi, and fire (Everham and Brokaw, 1996; Hély et al., 2000; Kafka et al., 2001; Pautasso et al., 2005; Griess and Knoke, 2011; Seidl et al., 2011b; Jactel et al., 2017), compared to remnant stands with homogeneous structure and composition.

Partially stand-replacing disturbances have some characteristics similar to SR, although to a lesser degree. With increasing size of the openings, surviving trees become more susceptible to subsequent mortality. Wind loading increases very rapidly with increasing size of the opening, up to a gap diameter of twice the height of the edge trees (Quine and Gardiner, 2007). Windthrow increases the proportion of edge habitats. New edges created by wind are susceptible to further wind damage due to sudden changes in wind loading that the trees are not acclimated to (Stathers et al., 1994; Gardiner et al., 1997; Peltola et al., 1999a). Zeng et al. (2010) found that when the proportion of gap areas increased from 10 to 20%, the length of edges at risk increased by 77–80% for different age classes. The total length of edges at risk reached a maximum when about 50% of the landscape was open gaps. The characteristics of the newly formed edges matter, as abruptness and density of the edges strongly influence wind loading (Stathers et al., 1994; Gardiner et al., 1997). Structural contrast between the forested and non-forested areas generally increases with increasing disturbance severity (Foster et al., 1998; Stueve et al., 2011), thus uneven sparse stand- and gap edges often resulting from moderate windthrows may be very vulnerable to further windthrow (Stathers et al., 1994; Gardiner et al., 1997).

Severity and extent of wind disturbance affect remnant stands’ vulnerability to bark beetle outbreaks: Schroeder and Lindelöw (2002) found that tree mortality caused by I. typographus peaked earlier in smaller stands with fewer wind-felled trees than in the larger stands with more wind-felled trees. They also found that the total number of wind-felled spruces and the area of the wind-disturbed forest were significantly correlated with the number of trees subsequently killed by I. typographus. Similarly, Kärvemo et al. (2014a) found that at the landscape level, the area of storm gaps was a significant factor affecting tree mortality from I. typographus. Sometimes after a moderate-severity disturbance, insect pests may exhibit a bimodal response—first they colonize the windthrown and dying trees and build up populations, then there may be some decline, but the abundance still stays at relatively high levels, and then the pest responds positively again as the remnant trees are further weakened and colonized over time (Vogt et al., 2020). It is also possible that medium-severity wind disturbances can result in higher beetle populations than high-severity events because extensive mortality can lead to abrupt collapse in suitable habitat provision when the dead trees become too dry for bark beetle colonization, whereas intermediate-severity disturbances can maintain suitable habitats longer (Stadelmann et al., 2013; Kärvemo et al., 2014a; Vogt et al., 2020). Moreover, the death of host trees exposes nonhost remnant trees to higher wind loads, making them more vulnerable to wind (Ruel et al., 2023).

Greenberg (2021) observed recovery dynamics following moderate-severity wind disturbance in a temperate upland hardwood forest. Formed gaps were 0.166–1.08 ha in size, an average of 24.0% of trees (41.1% BA) were windthrown, and 1.0% of trees (1.1% BA) died standing during wind disturbance within the gaps. Over the 21 years after the hurricane, 15.4% (16% BA) of trees additionally died in gaps, while mortality was 15.1% (14.1% BA) in the control plots. Although in gaps more trees died additionally due to windthrow (8.2% in gaps versus 3.1% in control), these long-term results indicate that overall delayed tree mortality was not substantially accelerated in gaps following the initial “pulse” of hurricane-related mortality. Similarly, Szwagrzyk et al. (2017) reported very low subsequent mortality 13 years after wind disturbance (plots with different severities pooled) in a mixed forest dominated by Scots pine; many quite heavily damaged (e.g., strongly bent or leaning) trees had survived. However, Sato et al. (2017) observed high delayed mortality following a moderate-severity typhoon disturbance in a natural mixed forest dominated by Sakhalin fir (Abies sachalinensis (F.Schmidt) Mast.). During a 7-year post-disturbance period, the annual mortality was approximately 2–4%, with averages of 3.5 and 2.8% during the periods of 1–2 years later and 3–7 years later, respectively, significantly exceeding the level recorded during the pre-disturbance period (average of 0.9%). The main tree species A. sachalinensis had especially high delayed mortality levels; during the 7-year post-disturbance period, additionally damaged basal area reached approximately 80% of that which occurred during the disturbance. Post-disturbance deaths of A. sachalinensis exhibited a spatial pattern of gradual expansion around A. sachalinensis trees that had died during the disturbance.

Several factors may account for large variations in mortality, including contingency—remnant trees may or may not experience another heavy wind. Similarly, weather conditions prior to, during, and after partial wind disturbance may or may not facilitate insect pests. Moreover, pest infestation and wildfire do not necessarily occur after wind disturbance, even if weather conditions favor such events (Szwagrzyk et al., 2017). Soil conditions (e.g., periodically waterlogged or hydromorphic soils) also add to the variation in mortality (Mitchell, 1995; Vašíčková et al., 2021).

On a single tree scale, the characteristics of a survivor such as species, age, and size largely affect its fate following PR (Table 4). The level of damage caused to its stem, roots and canopy during wind disturbance clearly influences its prospects as well (Webb, 1989; Everham and Brokaw, 1996; Arévalo et al., 2000). Arévalo et al. (2000) studied tree damage and mortality over 14 years following a moderate severity (approximately 50% mortality during the wind event; two forest types pooled) windthrow in permanent plots in an oak forest and a pine forest in central Minnesota. They found that the probability of a tree dying varied with respect to damage type and severity—mortality probability was approximately 0.18 for undamaged trees and 0.38 for lightly damaged trees. At the same time, for trees with broken stems the mortality probability increased to 0.84 and uprooted trees had only marginal chances (mortality probability 0.98) to survive throughout the study period. Interestingly, Vašíčková et al. (2021) noticed that storm Herwart (2017) selectively impacted the conifer population in the Czech Republic. They found that trees that germinated under the canopy and experienced several periods of suppression and release were more likely to survive the storm. Thus, the authors proposed that individuals with a more varied disturbance history with accompanying denser wood are more likely to survive recurring wind disturbances.

Disturbance of moderate severity might lower the incidence of diseases affecting relatively old trees for a given species. For example, balsam firs (Abies balsamea) in the southern boreal forest of North America become more susceptible to spruce budworm (Choristoneura fumiferana Clem.) attacks as they age due to declining concentrations of defensive secondary foliar compounds (Kumbașlı et al., 2011). Although other factors besides host age also influence population dynamics of C. fumiferana, windstorms that blow down older cohorts every few decades may reduce the incidence of this insect at stand- and landscape-scales.

Following FS, the affected stands are generally less vulnerable to subsequent wind disturbances as a result of the local high frequency-low intensity wind disturbance regime (McCarthy, 2001; Angelstam and Kuuluvainen, 2004). The relatively frequent winds enhance the adaptation of trees in the entire stand, including the trees at gap edges. For instance, Shorohova et al. (2009) concluded that 0.2–10.4% basal area removal by windthrow does not significantly increase subsequent mortality in boreal forests.

Runkle (1998) found that nearly half of the gaps (46%) expanded during a 14-year study period in a southern Appalachian old-growth hardwood-dominated forest. Mortality rates were higher for larger stems; American beech (Fagus grandifolia) exhibited greatest mortality rate (1.18% y⁻¹). The average mortality rate of canopy trees bordering gaps (0.60% y⁻¹) was, however, not significantly higher than canopy-tree mortality in general, suggesting that the presence of gaps did not influence tree mortality rates. Nevertheless, storms with higher wind speeds can cause significant expansion of gaps due to increased exposure to wind of trees at gap edges. Foster and Reiners (1986) found that stepwise gap expansion caused by windthrow and coalescence of gaps into larger gaps were common processes in the virgin subalpine forests located at Crawford Notch, New Hampshire. Bottero et al. (2011) also recognized gap expansion to be an important process in a temperate mixed old-growth forest in Bosnia and Herzegovina, whereas wind and snow were suggested to be the main agents of onward mortality. In another study from Bosnia and Herzegovina, Nagel and Svoboda (2008) found that 70% of gaps showed evidence of subsequent expansion after their initial formation in a mountainous forest reserve dominated by European beech (Fagus sylvatica L.) and silver fir (Abies alba Miller).

Pathogenic fungi are highly influential mortality agents in forests driven by gap dynamics (Hennon, 1995; Angelstam and Kuuluvainen, 2004). As infected trees fall, they often injure nearby trees, thereby creating entry points for new infections and contributing to the persistence of decay fungi in the stand (Hennon, 1995). Worrall et al. (2005) stated that gap expansion is a prominent feature of the disturbance regime of Picea rubens-Abies balsamea forests in the northern Appalachians mountains, as gap expansion was found to occur more frequently than gap initiation in these forests. The genus Abies was strongly associated with gap expansion; stem breakage under wind-loading was the most commonly identified onward mortality agent, whereas fungi causing root and butt rots were the most frequent biotic mortality agents, followed closely by spruce beetles (D. rufipennis).

The spatial and temporal dynamics of bark beetles and their natural enemies are linked to the disturbance regime. Forests that are primarily shaped by gap dynamics exhibit relatively constant and diverse supply of dead and dying trees (Bouget and Duelli, 2004; Vogt et al., 2020), ensuring the continuous presence of natural enemies and competitors of bark beetles (Bouget and Duelli, 2004). Nevertheless, the potential still exists for insect outbreak if there is a source of infestation nearby (Osetrov, 2002; Wermelinger, 2004; Kärvemo et al., 2014a), given that the stand contains suitable host trees. For instance, in Switzerland, a major bark beetle (mainly I. typographus) outbreak that started in 1993 in a 120-year-old Norway spruce stand presumably was a consequence of some scattered trees felled by storm Vivian. This outbreak culminated in almost total death of the stand by 1995 (Wermelinger, 2002). Described disturbance regimes and most likely subsequent disturbance agents associated with them are depicted in Figure 2. Summarized effects of windthrow on subsequent disturbances classified by landscape, stand and tree spatial scales are described in Tables 2–4.