Journal list menu

Volume 57, Issue 1 p. 67-76
RESEARCH ARTICLE
Free Access

Is salvage logging effectively dampening bark beetle outbreaks and preserving forest carbon stocks?

Laura Dobor

Laura Dobor

Faculty of Forestry and Wood Sciences, Czech University of Life Sciences Prague, Prague, Czech Republic

Search for more papers by this author
Tomáš Hlásny

Corresponding Author

Tomáš Hlásny

Faculty of Forestry and Wood Sciences, Czech University of Life Sciences Prague, Prague, Czech Republic

Correspondence

Tomáš Hlásny

Email: [email protected]

Search for more papers by this author
Werner Rammer

Werner Rammer

University of Natural Resources and Life Sciences (BOKU) Vienna, Vienna, Austria

Search for more papers by this author
Soňa Zimová

Soňa Zimová

Faculty of Forestry and Wood Sciences, Czech University of Life Sciences Prague, Prague, Czech Republic

Search for more papers by this author
Ivan Barka

Ivan Barka

National Forest Centre—Forest Research Institute Zvolen, Zvolen, Slovak Republic

Search for more papers by this author
Rupert Seidl

Rupert Seidl

University of Natural Resources and Life Sciences (BOKU) Vienna, Vienna, Austria

Search for more papers by this author
First published: 28 September 2019
Citations: 62

Abstract

en

  1. Salvage logging is one of most frequently applied management responses to forest disturbances world-wide. As forest disturbances are increasing, so too is the application of salvage logging, yet its effects on ecosystems remains incompletely understood. In the Norway spruce (Picea abies (L.) Karst.) forests of Europe, salvaging of windfelled trees is inter alia applied to reduce the risk of bark beetle outbreaks (mainly Ips typographus L.). By preventing further disturbances, salvage logging can conserve live tree carbon (C) in forest landscapes. At the same time salvage logging reduces C stocks in detrital pools via the extraction of disturbed trees, its net effect thus remains unclear.
  2. We used the forest landscape model iLand to explore the effect of a wide range of salvaging intensities on (a) subsequent bark beetle outbreaks, and (b) landscape-scale forest C stocks in a Norway spruce-dominated production forest in Slovakia under past and future climatic conditions.
  3. Climate change resulted in a two- to three-fold increase in bark beetle disturbances throughout the 21st century in our simulations. We found that removing >95% of disturbed trees can effectively buffer the effect of increasing disturbances, dampening bark beetle infestations and increasing live tree C. Total ecosystem C followed a U-shaped pattern over salvaging intensity, with highest values in no salvage and 100% salvage scenarios.
  4. However, realistic rates of salvaging (<95% of disturbed trees detected and removed) had no significant effect on bark beetle dynamics and live tree C, and reduced the total ecosystem C stored in the landscape. Furthermore, the effect of reduced bark beetle disturbance under intensive salvaging was partly offset by increased wind disturbance.
  5. Synthesis and applications. Clearing disturbed areas to prevent future disturbances from bark beetles and conserve live tree carbon should only be applied where very high salvaging rates are feasible (i.e. small and concentrated disturbances). Considering that changing disturbance regimes make high-intensity salvaging increasingly challenging, alternative disturbance management approaches need to be developed.

Foreign Language Abstract Czech

cs

  1. Odstraňování napadených či poškozených stromů (tzv. nahodilé těžby) je nejrozšířenější reakcí na výskyt různých disturbancí v lesních ekosystémech. Klimatická změna vede k intenzívnějšímu vlivu disturbancí na lesy a je tedy zapotřebí očekávat i nárůst rozsahu a intenzity nahodilých těžeb. Ve smrkových porostech v Evropě se nahodilé (v tomto případě přesněji sanitární) těžby větrem poškozených stromů provádějí proto, aby se zabránilo náletu lýkožroutů na vhodný rozmnožovací materiál či snížil nárůst populace, která již polámané stromy obsadila. Pokud zabráníme přemnožením kůrovců, uchováme rovněž zásobu uhlíku v živých stromech. Na druhé straně, těžbami odstraňujeme část uhlíku uloženého v mrtvém dřevu. Výsledný efekt těchto dvou protichůdných procesů na celkovou uhlíkovou bilanci však není znám.
  2. V této studii jsme pomocí ekosystémového modelu iLand hodnotili vliv různých intenzit odstraňování větrem poškozených stromů na (a) následné přemnožení kůrovce a (b) na celkovou zásobu uhlíku v ekosystému. Výzkum byl realizován v hospodářském lese s dominancí smrku (75%) s rozlohou 16 050 ha.
  3. V důsledku změny klimatu došlo v průběhu 21. století k dvou až trojnásobnému nárůstů objemu stromů napadených kůrovcem. Odstranění více než 95% větrem poškozených stromů účinně zmírnilo intenzitu kůrovcových přemnožení a došlo tím ke zvýšení zásob uhlíku v živých stromech. Závislost celkové zásoby uhlíku v ekosystémů na různé intenzitě těžeb měla tvar U. Nejvyšší úrovně byly dosahovány při absenci těžeb (uhlík se více kumuluje v mrtvém dřevu) a při celkovém odstranění větrem poškozených stromů (kůrovec je potlačen a uhlík se kumuluje v živých stromech). Realistická intenzita sanitárních těžeb (detekce a odstranění <95% poškozených stromů) neměla na dynamiku kůrovce a zásobu uhlíku v živých stromech téměř žádný vliv, naopak způsobila pokles celkové zásoby uhlíku v ekosystému. Intenzivní prevence vzniku přemnožení kůrovce zároveň vyvolala v dlouhodobém horizontu nárůst poškození porostů větrem a částečně tak negovala snahu o prevenci disturbancí.
  4. Z těchto výsledků vyplývá, že odstraňování větrem poškozených stromů s cílem prevence nebo potlačení přemnožení kůrovce a uchování zásob uhlíku má smysl realizovat pouze tehdy, pokud je možné dosáhnout velice vysokou intenzitu těžeb (např. při menších a soustředěných polomech). Protože s klimatickou změnou dochází k zesílení vlivu disturbancí a změně dlouhodobých disturbančních režimů, provádění vysokých intenzit sanitárních těžeb bude stále obtížnější. Je proto zapotřebí hledat alternativní způsoby managementu hospodářských lesů orientující se na podporu jejich přirozené rezilience.

1 INTRODUCTION

Stand replacing forest disturbances such as wind, insect outbreaks or fires and the management responses they trigger are among the most severe perturbations of ecosystem processes (Leverkus, Rey Benayas, et al., 2018; Lindroth et al., 2009; Seidl, Schelhaas, Rammer, & Verkerk, 2014). Disturbances are natural processes in forest ecosystems and are important drivers of ecosystem dynamics (Stephens et al., 2013; Thom, Rammer, Dirnböck, et al., 2017; Thom, Rammer, & Seidl, 2017a). They often facilitate biodiversity (Thom & Seidl, 2016), mitigate forest vulnerability to future disturbances (Seidl, Donato, Raffa, & Turner, 2016; Stephens et al., 2013), and foster autonomous adaptation of forests to rapidly changing environmental conditions (Kulakowski et al., 2017; Lamers, Junginger, Dymond, & Faaij, 2014; Thom et al., 2017; Thom, Rammer, & Seidl, 2017a). However, disturbances also alter biogeochemical cycles in ecosystems, including carbon (C), nutrients and water and reset natural forest succession (Dobor et al., 2018; Mikkelson, Dickenson, Maxwell, McCray, & Sharp, 2013; Seidl, Rammer, & Spies, 2014). Consequently, disturbances are often perceived negatively in managed forest because they compromise the provisioning of ecosystem services and can have negative impacts on the economy of communities depending strongly on revenue from forests (Rosenberger, Bell, Champ, & White, 2013). Humans have therefore applied strategies to mitigate the risk of disturbances for many decades (Roberge et al., 2016; Seidl, 2014). However, long-term disturbance prevention is not consistent with natural ecosystem dynamics (Holling & Meffe, 1996). Furthermore, efforts to prevent risk in the short term can result in increased mid- to long-term risks, for example via creating a high share of overmatured stands, high biomass stocks or simplified vertical and horizontal structure (Reyer et al., 2017).

Salvage logging is one of most widespread management responses to forest disturbances world-wide (Leverkus, Lindenmayer, Thorn, & Gustafsson, 2018; Leverkus, Rey Benayas, et al., 2018; Müller et al., 2018). Salvage logging is the felling and removal of trees in naturally disturbed forests with the primary intention to recoup economic losses, reduce hazards to infrastructure and ensure human safety (Molinas-Gonzáles, Leverkus, Marañón-Jimenéz, & Castro, 2017). Furthermore, in many ecosystems salvage logging also aims to fulfil a sanitary role, by reducing the risk from subsequent disturbances. In Norway spruce (Picea abies (L.) Karst.) forests, for instance salvage logging of windfelled timber is frequently aimed at removing broken or uprooted trees which serve as breeding substrate for native bark beetles (mainly Ips typographus L.) (Schroeder, 2007; Stadelmann, Bugmann, Meier, Wermelinger, & Bigler, 2013). Bark beetles swiftly colonize such trees because of their weakened defences (Komonen, Schroeder, & Weslien, 2011), and the corresponding population build-up triggers a transition from endemic to epidemic population dynamics (Kausrud et al., 2012), with beetles spreading into healthy forests around windthrown areas. This interaction effect can trigger large outbreaks of bark beetles, which—under favourable conditions—can substantially exceed the extent of the initial wind disturbance (Mezei et al., 2017; Økland, Nikolov, Krokene, & Vakula, 2016).

Recent increases in forest disturbances (Seidl, Schelhaas, et al., 2014; Senf et al., 2018) have led to an unprecedented increase in salvage logging (Leverkus, Lindenmayer, et al., 2018). However, salvage logging has tangible effects on biodiversity (Thorn et al., 2017) and can compromise mechanisms underlying ecosystem resilience (Ghazoul, Burivalova, Garcia-Ulloa, & King, 2015). Furthermore, concerns have emerged that salvage logging may affect cultural, regulating and supporting ecosystem services throughout the ecosystem services cascade (Leverkus, Rey Benayas, et al., 2018). In Europe`s spruce forests, one prominent example in the context of climate regulation is the effect of salvage logging on forest carbon (C) storage. Salvage logging reduces forest C storage by removing the C stored in disturbed trees, but can also enhance live tree C pools by dampening bark beetle outbreaks. Due to the complex interplay between these effects the impact of salvage logging on forest C stocks remains unclear. Moreover, salvaging affects multiple ecosystem processes and their interactions (Leverkus, Rey Benayas, et al., 2018), potentially resulting in non-additive outcomes. This is of particular concern in the context of climate change, as future outcomes of salvage logging could differ from the past due to nonlinear effects of changing environmental conditions. Climate change can, for instance increase the severity of bark beetle disturbances in their native range, facilitate the expansion of beetle populations to new locations (Cudmore, Björklund, Carroll, & Lindgren, 2010), and amplify the interactions between wind and bark beetles (Seidl & Rammer, 2017). However, climate change can also create negative feedbacks via favouring warm-adapted broadleaved species and reducing the proportion of host tree species of bark beetles (Temperli, Bugmann, & Elkin, 2013; Thom, Rammer, Dirnböck, et al., 2017; Thom, Rammer, & Seidl, 2017b).

Here we used the individual-based forest landscape and disturbance model iLand (Seidl, Rammer, Scheller, & Spies, 2012) to investigate how different salvaging intensities of windfelled trees affect bark beetle disturbances, and to quantify disturbance impacts on landscape-scale C storage. We specifically addressed the trade-offs between C removal by salvaging and C increases by mitigated bark beetle disturbances, asking whether there is an optimal salvaging intensity balancing the positive and negative effects on forest C storage. Finally, we investigated how climate change—affecting both forest growth and disturbance regimes—impacts ecosystem C storage, and whether the effect of salvage logging is modulated by changing climatic conditions. We focused our analyses on a managed temperate forest landscape in Central Europe which has experienced severe wind and bark beetle disturbances followed by intensive salvaging operations in recent years.

2 MATERIALS AND METHODS

2.1 Study region

The Goat Backs Mts. study region is located in central-eastern Slovakia (Lon 20.088—20.275, Lat 48.920—49.061) and covers an area of 16 050 ha (Figure 1). The forest cover is 70% and is dominated by Norway spruce (Picea abies (L.) Karst.), which makes up 75% of the forest area. The remaining species are European larch (Larix decidua Mill.; 10%), Scots pine (Pinus sylvestris L.; 9%), Silver fir (Abies alba Mill.; 3%) and European beech (Fagus sylvatica L.; 2%). The elevation range is 620–1550 m a. s. l. The annual mean air temperature during the growing season (April–September) in the period 1996–2016 was 12°C, and growing-season precipitation was 692 mm. Cambisols and Podsols prevail, whereas Rendzinas occur on calcareous bedrock, which forms the bedrock in the highest reaches of the landscape.

Details are in the caption following the image
Map of the study region and its location in Central Europe. Forest area and the extent of the recent wind and bark beetle disturbance episode (2007–2016) were derived from Landsat satellite imagery. The inset photograph illustrates the situation on site, showing recently bark beetle killed trees (background) as well as indications of recent salvage harvesting (stumps in the foreground)

The study region is owned by the church and is under the stewardship of a private forest management enterprise. An even-aged management system with a rotation period of approximately 100 years is applied throughout the landscape. The dominant silvicultural approach to tree regeneration in stands with fir and/or beech admixtures is a uniform shelterwood cut. In the dominant Norway spruce monocultures, a clearcut system is applied. Maximum cutblock size is 3.0 ha in all systems.

The landscape and its recent disturbance history is typical for many Central European forests, and is characterized by severe large-scale disturbances in recent years (Senf et al., 2018). Specifically, the study region has experienced intensive wind and bark beetle (mainly Ips typographus L.) disturbances from 2007 onward, which affected as much as 39% of the regional forests (Dobor et al., 2018). The landscape level growing stock decreased from 4.22 Mill. m3 in 1996 (average of 380 m3/ha) to 1.93 Mill. m3 in 2016 (average of 173 m3/ha). At the same time, the landscape-level forest age distribution was substantially shifted towards an overabundance of young stands (Source: forest management plans [FMP]; National Forest Centre, Slovakia). The main management response to this recent disturbance episode has been a high level of salvage logging, and a reduction in planned harvests. Salvage logging was generally applied with high intensity, although a portion of the disturbed trees remained on site for more than one season for logistical reasons. Mass use of pheromone traps was another measure applied to monitor and reduce bark beetle populations.

2.2 Simulation model

We used the individual-based forest landscape and disturbance model iLand (Seidl, Rammer, et al., 2012) to dynamically simulate wind and bark beetle disturbance in the study landscape, and evaluate the response of the forest C cycle to various salvaging intensities under climate change. The main entities simulated in iLand are trees, for which the demographic processes of growth, mortality and regeneration are simulated. Processes at the stand and landscape-scale constrain the dynamics of individual trees, and large-scale patterns emerge from tree-level interactions (Seidl, Rammer, et al., 2012).

Trees are simulated as adaptive agents that compete for resources (light, water and nutrients) (Seidl, Rammer, et al., 2012). Production physiology is modelled in a simplified process-based manner using a light use efficiency approach (Landsberg & Waring, 1997). Species-specific environmental modifiers are applied to account for the effect of environment on the total intercepted radiation (APAR), which drives gross primary production (GPP). Temperature, soil water availability, vapour pressure deficit, soil nitrogen availability and atmospheric CO2 are considered influences on GPP. APAR further depends on tree leaf area, tree position in the canopy and radiation use strategy (light-demanding or shade-tolerant). Individual tree mortality is simulated based on species-specific maximum size and age as well as the occurrence of stress (Seidl, Rammer, et al., 2012). C starvation is used as a process-oriented indicator of tree stress, and can result from competition for resources as well as suboptimal environmental conditions for tree growth (e.g. drought). The model simulates live C stocks in stem, branch, foliage, coarse root and fine root compartments. Snags and the transition from snags to downed woody debris are considered explicitly. A closed C cycle is simulated by also tracking the fate of C in detritus and soil pools (Thom, Rammer, Dirnböck, et al., 2017; Thom, Rammer, & Seidl, 2017b).

iLand simulates disturbances in a spatially explicit manner and currently contains process-based modules for wind (Seidl, Rammer, & Blennow, 2014), bark beetle (Seidl & Rammer, 2017) and fire disturbances (Seidl, Rammer, & Spies, 2014). Wind disturbances are simulated based on wind data such as peak wind speed, wind direction and storm duration. The model initiates wind disturbance in locations where canopy rugosity changes abruptly, that is where vertical differences between the top heights of neighbouring grid cells exceed 10 m (e.g. Blennow & Sallnäs, 2004). Wind speed at the canopy top height is calculated based on a vertical wind profile at the stand edge. Wind data and individual tree turning coefficients are used to calculate critical wind speeds for uprooting and tree breakage based on the approach of Gardiner, Peltola, and Kellomäki (2000). The effect of a given wind event is simulated iteratively, with forest structure (including the appearance of new edges) being updated after each iteration (horizontal resolution: 10 m grid cells). Simulated wind disturbances thus emerge dynamically from the interplay of landscape structure and configuration with a given wind event.

With regard to bark beetle disturbances iLand considers bark beetle phenology and development, the spatially explicit dispersal of beetles, host tree colonization and defence, as well as temperature-related overwintering success (Seidl & Rammer, 2017). Host trees are Norway spruce trees with a diameter at breast height (DBH) of >15 cm. An outbreak can be triggered by a wind disturbance simulated by the model, or occurs based on a climate-sensitive background probability. Bark beetle development is simulated based on the climate-sensitive development rates for each development stage (Baier, Pennerstorfer, & Schopf, 2007). The model tracks beetle cohorts rather than individuals. The cohort is defined as the minimum number of beetles needed to colonize a tree. Every brood tree disperses a number of beetle cohorts determined by the reproductive rate of the beetle, estimated to range between 4 and 24 (Wermelinger & Seifert, 1999). The dispersal of beetles is simulated in a two-stage approach: First, a symmetrical dispersal kernel is used to calculate the approximate flight distance. Second, beetles actively search for suitable host trees in the local neighbourhood determined via the dispersal kernel (Kautz, Schopf, & Imron, 2014). In this search, wind-disturbed trees are being preferred over healthy trees. For healthy trees, attacking beetle cohorts need to overcome a tree's defence system, which is approximated by its non-structural carbohydrate reserves. Trees can be attacked by multiple waves of beetles per year if the climate allows for the development of more than one beetle generation.

The model was extensively tested across a range of ecosystems in Europe and North America in previous studies (Seidl, Spies, et al., 2012; Silva Pedro, Rammer, & Seidl, 2015; Thom, Rammer, Dirnböck, et al., 2017). Extensive testing of productivity, natural mortality and regeneration patterns for the Goat Back Mts. study landscape was conducted by Dobor et al. (2018). All tests showed satisfactory performance of the model in this study region (see also Appendix S1).

2.3 Landscape initialization and climatic drivers

We initialized the landscape based on data from FMP provided by the National Forest Centre of Slovakia. The data are collected in the field in 10-year cycles and are recorded for forest stands, which are polygons with a variable size seamlessly covering the total forest area of the landscape. The attributes used to initialize the landscape were tree species, number of trees per hectare, stand age and DBH. Individual tree diameters were randomly drawn from diameter distributions centred on the mean DBH of each stand, and tree height (H) was estimated based on the regional DBH:H functions. Trees below 4 m height were initialized as height cohort (with representative individuals describing groups of similar-sized trees) based on stand-level information derived from FMP.

Soil depth and nitrogen information required to run iLand were derived on a 100 × 100 m grid from the national forest soil database (National Forest Centre). Daily meteorological data from the nearby meteorological station were used for deriving the climate time series driving the simulations (see Dobor et al., 2018 for details). Regional climate model (RCM) simulations conducted in the framework of the CORDEX project (Giorgi, Jones, & Asrar, 2009) were used to evaluate the effect of climate change. Six GCM-RCM combinations driven by two Representative Concentration Pathway (RCP) scenarios (RCP4.5 and RCP8.5) were used. More details on used climate change scenarios can be found in Appendix S2.

2.4 Experimental design

Simulations were run for the period 1997–2100, assuming a continuation of current forest management. The simulated management operations included planting, tending, thinning and harvesting, with timing and intensity of operations modelled after the management practice currently applied in the region (see Dobor et al., 2018 for more details). The incidence of disturbances and subsequent salvage logging supersede regular management operations, resetting the default stand treatment program.

Seven different salvaging intensity scenarios were tested, corresponding to 0, 20, 40, 60, 80, 95 and 100% of trees salvaged. Both trees killed by wind and bark beetles were salvaged in the year of disturbance. No preventive removal of live trees aiming to halt the spread of bark beetles (sanitation logging) was performed. Most model parameters driving bark beetle dynamics were kept default as reported in Seidl and Rammer (2017) (see Appendix S3 for the parameter values used).

As the projection of extreme wind events in climate models is still highly uncertain we generated time series of future storm events based on past observations. We assumed five wind events to occur between 1997 and 2100, set at the simulation years 5, 30, 50, 75 and 90. For each of these wind events, we derived wind speeds by drawing from a Gumbel distribution parameterized based on the data observed in the nearby meteorological station (see Appendix S4), and set the wind duration to 90 min. In order to account for the stochasticity in future wind events we generated five future wind scenarios from these distributions. Overall, the simulated average wind damage of 0.9–1.2 m3 ha−1 year−1 (range of five simulated wind scenarios) corresponds well with the average wind disturbance rate of 0.88 m3 ha−1 year−1 (inter-annual range 0.35–2.56 m3 ha−1 year−1) reported for Slovakia for the period 1990–2015 (Konôpka, Zach, & Kulfan, 2016). The total number of simulations conducted was 455, consisting of 7 salvaging intensities × 5 wind scenarios × 13 climate scenarios (2 × RCPs × 6 models + reference climate).

Using this simulation framework, we evaluated the effect of salvage logging on total and live landscape C, amounts of C in trees affected by wind and bark beetle disturbance, and the interactions between these variables.

3 RESULTS

3.1 Salvaging effects on bark beetle disturbance

In all simulations, the amount of C in Norway spruce trees killed by bark beetles was strongly affected by salvaging intensity (Figure 2a). The complete removal of windblown and bark beetle killed trees (SI 100%) prevented bark beetle outbreaks, and the amount of C in killed spruce trees was low under reference climate in this salvaging scenario (see also Appendix S5). Climate change increased bark beetle disturbances even under a complete removal of windblown trees, and beetles were able to kill 0.2–0.4 tC ha−1 year−1 (average over the simulation period, including non-outbreak years).

Details are in the caption following the image
The effect of salvage logging intensity (% of disturbed trees removed) on the live tree C affected by bark beetles (tC ha−1 year−1, a), and the average live tree C stock of Norway spruce on the landscape (tC/ha, b) for different climate scenarios. Each Representative Concentration Pathway (RCP) storyline consists of six different climate model combinations, and each variant was driven by five different wind scenarios. Annual averages over a 104-year simulation period (1997–2100) are presented. Boxes represent the inter-quartile range and whiskers extend to the minimum and maximum values

The amount of live tree C affected by bark beetles increased strongly nonlinearly with decreasing salvaging intensity. The retention of only a minor amount of windblown trees (salvaging intensity 95%, which resulted in the retention of 0.07–0.11 tC/ha in dead trees on site; Appendix S6) increased bark beetle disturbances substantially, reaching 0.3 tC ha−1 year−1 under reference climate (Figure 2). For even lower salvaging intensities (80, 40, 60 and 20%) the annual amount of live tree C affected by bark beetles was in the range 0.4–0.5 tC ha−1 year−1. These findings indicate that salvaging intensities below 80% have very little effect on bark beetle disturbances.

Climate change increased the amount of live tree C affected by bark beetles two to threefold. Consequently, it also decreased the amount of Norway spruce live tree C (Figure 2b). Climate change also resulted in greater volumes of salvaged C (Appendix S6). While the highest salvaging intensity under reference climate removed 1.2 tC ha−1 year−1, the removal by salvage harvesting under climate change reached as much as 1.7 tC ha−1 year−1 (average over the simulation period). The differences between the RCP scenarios were minor, although RCP8.5 simulations showed consistently greater bark beetle influence than those under RCP4.5.

3.2 Salvaging effects on landscape carbon

Live tree C on the landscape benefited from salvage logging and increased by 6%, from 149 tC/ha (no salvaging) to 159 tC/ha (100% of disturbed trees salvaged) under reference climate (Figure 3a). The effect of live tree C saved by salvaging was even more pronounced under climate change, and was 8% under both RCP scenarios. The overall climate response of Clive regardless of salvaging intensity was variable, with RCP8.5 resulting in increased and RCP4.5 in decreased live tree C compared to the reference climate.

Details are in the caption following the image
Response of live tree carbon (a) and total ecosystem carbon (b) in the landscape (t/ha) to different salvaging intensities under different climate scenarios. Each RCP storyline consists of six different climate model combinations, and each variant was driven by five different wind scenarios. The values are averages over a 104-year simulation period

In contrast, total ecosystem C decreased with increasing salvage intensity. Consequently, the effect of C removal via salvaging was stronger than the dampening effect of salvaging on bark beetle disturbances (Figure 3b). At very high salvaging intensities (i.e. >95%), however, Ctotal increased again, with strongly reduced mortality compensating for losses from timber extraction. The strongest negative effects of salvaging on Ctotal were thus for SIs of 80%–95%, while the difference between SIs of 0 and 100% were minor (between 408 and 404 tC/ha). The effect of climate change on Ctotal was positive, with total ecosystem C storage increasing by 3% and 3.5% under RCP4.5 and RCP8.5 respectively (from 405 to 416 and 419 tC/ha).

3.3 Interactions between wind and bark beetle disturbances

Higher salvaging intensities resulted in higher levels of live tree C affected by wind disturbance (Figure 4a). This effect was minor under reference climate but was substantial under climate change. Climate change generally decreased the amount of live tree C affected by wind disturbance by 0.18–0.23 tC ha−1 year−1 for SIs up to 80%, with lower effects under higher SIs.

Details are in the caption following the image
The amount of live tree carbon affected by wind disturbance under different salvaging intensities (a) and relationship between live tree C affected by wind and bark beetles (b). The results reached under reference climate and two groups of climate change scenarios (RCP4.5 and RCP8.5) are shown. Linear fit and prediction interval are shown

A considerable trade-off was evident also between wind and bark beetle disturbances. Live tree C affected by bark beetles was negatively correlated with live tree C affected by wind (Figure 4b). The increase in bark beetle disturbances in response to climate change (from 0.38 to 0.89 tC ha−1 year−1; see also Figure 2) substantially decreased wind disturbance (from 0.89 to 0.73 tC ha−1 year−1). This indicates that the efforts to reduce bark beetle disturbance by salvaging can—in part—be compromised by an increased forest susceptibility to wind. However, an increased impact of bark beetles due to climate change may result in a concomitant decrease in wind impact.

4 DISCUSSION

Salvage logging is the most frequent management response to forest disturbances world-wide, but its effects on forest ecosystems remain incompletely understood. In the forests of Central Europe salvaging has been applied at unprecedented rates in recent years in response to increasing natural disturbances. In these systems, salvaging of windfelled trees is inter alia practiced to reduce the availability of breeding substrate for bark beetles, yet the efficacy of the measure in influencing bark beetle dynamics remains poorly documented. We here quantified the influence of salvaging on subsequent bark beetle dynamics, and estimated the landscape-scale C storage effect of different levels of salvage logging for a temperate forest landscape in Europe.

4.1 Salvage logging as a means for preventing bark beetle outbreaks

The simulations showed that only the highest salvaging intensities reduced bark beetle disturbance and effectively preserved Norway spruce live tree C. Where the retention of high levels of Norway spruce is of key concern, a very high level of salvage logging is a potent management measure. This finding is consistent with recent management applications of salvaging in Central Europe, aiming to reduce the amount of breeding substrate for bark beetles and thus reducing subsequent bark beetle disturbances (Hlásny & Turčáni, 2013; Økland et al., 2016). However, positive salvaging effects sharply decreased for salvaging intensities of <95%, indicating that even retaining a small amount of wind-felled spruce trees is sufficient for the beetle to make the critical transition from endemic to epidemic population dynamics (Kausrud et al., 2012). Yet, reaching the high salvaging intensities required for positive salvage effects may not be feasible in forest management, particularly when windthrown areas are large and scattered. In contrast to current practice this insight suggests that the removal of windfelled trees with the intention to prevent bark beetle outbreaks should only be applied when such a removal can be achieved with a very high intensity (e.g. when windfelled areas are small and accessible). It is important to note, however, that we here did not analyse the ability of salvage logging to prevent a particular disturbance from spreading, but rather focused on the long-term effects over time frames of typical rotation periods. Moreover, our findings do not address other motivations for salvage logging, such as the recuperation of economic losses or the protection of infrastructure from falling trees (Leverkus, Lindenmayer, et al., 2018; Molinas-Gonzáles et al., 2017).

Climate change is amplifying bark beetle disturbance (Cudmore et al., 2010; Seidl & Rammer, 2017) via weakening tree defence, increasing insect reproduction and reducing winter mortality of bark beetles (Appendix S7). In our simulations, the amount of live tree C affected by bark beetles increased two- to threefold in response to climate change, which is consistent with, for example expected increases for the European Alps (Seidl, Schelhaas, Lindner, & Lexer, 2009). High-intensity salvage logging reduced the amount of live tree C affected by bark beetles also under climate change. Specifically, our results indicate that if a 100% salvage level could be achieved, climate-induced increases in bark beetle disturbances could be efficiently prevented. For realistic levels of salvaging (i.e. <100%), however, bark beetle disturbances increased strongly in our simulations. This indicates that management responses that were deemed successful in the past, such as salvage logging, are not efficient in preventing a climate-related intensification of disturbance regimes. Measures reducing disturbance risk and increasing forest resilience to disturbance are thus needed (Seidl, 2014). Such measures include silvicultural approaches creating forests that are less prone to the impact of natural disturbances (Seidl, Albrich, Thom, & Rammer, 2018), and that facilitate fast recovery from disturbances once they occur (Johnstone et al., 2016). Specifically, mixed species forests containing both early- and late-seral species and structurally diverse stands with a layer of advanced regeneration in the understorey should be promoted. Retaining biological legacies such as surviving trees and parts of deadwood is another potent means to foster forest resilience to disturbances (e.g. Castro et al., 2011).

4.2 Effects of salvage logging on landscape carbon

Total landscape C decreased only moderately with increasing salvaging intensities. This weak response underlines the trade-offs between C removed via salvaging and live tree C saved from subsequent disturbances. Similarly, insignificant effects of salvaging on Ctotal were found by Bradford et al. (2012) in a boreal forest disturbed by wind and fire. Moreover, the effects of climate change on total ecosystem C storage were considerably stronger than the effects of salvaging in our simulations, indicating that drivers beyond the influence of forest management might become more important in the future (Seidl et al., 2019). This suggests for management to focus resources on creating disturbance-resilient forests rather than on short-term disturbance prevention.

We here focused on C storage in situ, yet forests also influence the climate via C storage in wood products, the substitution of fossil-based resources as well as changes in forest albedo and latent heat flux (Canadell & Raupach, 2008; Valsta et al., 2017). C in salvaged timber, for instance constitutes a lateral C flux that eventually results in C storage in wood products (e.g. Lamers et al., 2014). As salvage logging potentially influences climate regulation more broadly than analysed here, future work could, for example include the effect of C storage in wood products pools, in order to increase the comprehensiveness of our understanding of salvage logging effects.

In contrast to total ecosystem C, live tree C benefited from salvaging, increasing by between 6% and 8% under highest salvaging intensity. The effect of climate change on Clive varied with RCP scenario. A likely reason is the elevated atmospheric CO2 concentration under RCP8.5, which results in higher C uptake rates in this scenario family. This is in line with the previously reported sensitivity of post-disturbance forest recovery to the increased levels of CO2 (Dobor et al., 2018). The persistence of a CO2 fertilization effect, however, remains debated in the literature (Hararuk, Campbell, Antos, & Parish, 2019; Reyer et al., 2014) and is strongly moderated by mycorrhizal associations (Terrer, Vicca, Hungate, Phillips, & Prentice, 2016).

4.3 Effects of salvage logging on disturbance interactions

Many ecological effects of salvaging are thought to emerge from modified interactions in forest ecosystems (Leverkus, Lindenmayer, et al., 2018). This notion was strongly supported by our quantitative simulation results. Specifically, we observed two types of disturbance interactions being altered by salvage logging—the interaction modification and the interaction chain (Foster, Sato, Lindenmayer, & Barton, 2016).

The direct interaction between wind and bark beetle disturbances in the simulations was mediated by the presence of freshly windfelled trees. Salvaging modified this interaction by reducing the amount of such trees. As expected, this reduction mitigates the impact of bark beetle disturbance. In this regard our findings are well in line with observations (e.g. Økland et al., 2016; Stadelmann et al., 2013). We, however, also found this effect to be strongly nonlinear, with very high salvaging intensities being required in order to achieve this interaction modification.

A more complex and hitherto unrecognized effect emerged from the long-term interaction chain between wind and bark beetles. We found that while high-intensity salvaging can reduce the impacts of bark beetle outbreaks, such a reduction is partly compensated (but not cancelled out) by increased forest susceptibility to wind disturbance. The main mechanism behind this compensation was that mature Norway spruce stands were most susceptible to both wind and bark beetle disturbances (Hlásny & Turčáni, 2013; Wermelinger, 2004). Specifically, the intense suppression of bark beetle disturbances increased the share of dense and mature forests on the landscape, which are subsequently more susceptible to the wind disturbance (Jactel et al., 2009). Wind and bark beetles are thus ‘competing’ for the most vulnerable stands, and reducing bark beetle outbreaks retains more susceptible stands to be affected by wind.

More broadly, this finding indicates that a strong focus on disturbance prevention may generate overly vulnerable conditions with high levels of C stocks that are susceptible to a diverse set of hazards. In our simulations the reduction of bark beetle disturbances via salvaging led to the opposite effect of the one intended with regard to wind disturbance; a phenomenon known in wildfire management as the firefighting trap (Collins, Neufville, Claro, Oliveira, & Pacheco, 2013). Similar reasons have, for example led to a change in the attitude to wildfire management in the USA (Stephens et al., 2013). We suggest that future research should focus on a balanced approach of reducing risks and fostering resilience (Seidl, 2014), rather than aiming to minimize a single risk while inadvertently increasing others.

5 CONCLUSIONS

Salvaging of trees killed by wind and bark beetles is extensively applied in the coniferous forests of Europe. Our simulations indicate that if the aim of salvage harvesting is to dampen future bark beetle disturbances and conserve live tree C, very high salvaging intensities (i.e. >95% of disturbed trees detected and removed) need to be applied. The application of salvage harvesting is primarily recommended for small and concentrated disturbances where very high salvaging intensities are feasible in practice. Furthermore, our results highlight unexpected compensatory effects, such as increased wind disturbances in response to reduced bark beetle disturbances. We thus conclude that novel management responses to changing forest disturbance regimes are needed, going beyond disturbance prevention and focusing on disturbance resilience.

ACKNOWLEDGEMENTS

This study was supported by the grants ‘EVA4.0', No. CZ.02.1.01/0.0/0.0/16_019/0000803 financed by OP RDE; and projects of the Ministry of Education, Science, Research and Sport of the Slovak Republic under contracts no. APVV-15-04-13 and APVV-16-0325. R. Seidl and W. Rammer further acknowledge support from Austrian Science Fund FWF START grant no. Y895-B25. We thank two anonymous reviewers for helpful comments on an earlier version of the manuscript.

    AUTHORS' CONTRIBUTIONS

    L.D. and W.R. conducted all simulations and data analyses, T.H. and R.S. conceived the ideas and led the writing of the manuscript, I.B. and S.Z. collected the data. All authors contributed critically to the drafts and gave final approval for publication.

    DATA AVAILABILITY STATEMENT

    Results are archived in Zenodo open-access repository, https://doi.org/10.5281/zenodo.3244166 (Dobor et al., 2019). Additional information on the used ecosystem model can be found at http://iland.boku.ac.at.