The Impact of Salvage Logging on the Supply of Roundwood Assortments from Austrian Forests – a Statistical Analysis

Schlüsselbegriffe: Rohholz Sortimente, Endnutzung, Durchforstung, Holzeinschlagsmeldungen, Sturmschäden, Borkenkäferschäden, Schnee­druck, Störungen

Available at https://doi.org/10.53203/fs.2504.3

Download as PDF

See below the issue 4/2025 as E-Paper or have a look at our E-Paper archive dating back to 1955.

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

Abstract

This study analyses whether and how the amount of total timber from salvage logging has an impact on the shares of assortments of domestic timber supply as well as on the relationship between thinnings and final cuts in Austria. The main data source for this study are the annual „Timber Felling Reports“ of the (currently labeled) „Federal Ministry Agriculture and Forests, Climate and Environment Protection, Regions and Water Management“. Correlation analysis and analysis of variance (ANOVA) are the main statistical methods used, based on the time-series data 2006-2023. The results show, that the annual fluctuations of total timber from salvage logging are much more volatile than those of the assortments, the shares of which harvests are only moderately affected by the amount of salvage logging. For coniferous as well as non-coniferous wood, the share of sawlogs is not significantly decreasing and the shares of pulpwood and wood for energy are not significantly increasing when storm damage is increasing; rather it is the other way around. When the damage caused by other reasons than storm (e.g. bark beetles, snow-break) is increasing, then there generally is a decrease in the share of sawlogs. For coniferous wood the impact of damage other than storm is mainly and (partly) significantly positive on the supply of wood for energy, while for non-coniferous wood the main and (partly) significantly positive impact is on pulpwood supply. The supply of timber from salvage logging is above the average in final cuts (larger dimensions) and below average in thinnings (smaller dimensions). The correlation analyses show that increasing calamities affect final cuts more than thinnings, therefore the supply of sawlogs in general tends to increase slightly. Although the underlying data on timber harvests allow detailed analyses due to the differentiation by ownership categories, assortments, coniferous and non-coniferous timber as well as reasons for the damage (e.g. storms, bark beetles), the interpretation of the results is limited, because the data situation does not allow the inclusion of assortment qualities and assortment prices. In addition, the length of the time-series (2006-2023) is quite short. Several suggestions for potential follow-up research activities are presented. 

Zusammenfassung

In dieser Studie wird untersucht, ob und wie sich die Menge an Schadholz auf die Sortimentsanteile des heimischen Holzangebots sowie auf das Verhältnis zwischen Vor- und Endnutzung in Österreich auswirkt. Hauptdatenquelle für diese Studie sind die jährlichen „Holzeinschlagsmeldungen“ des (aktuell benannt) „Bundesministerium Land- und Forstwirtschaft, Klima- und Umweltschutz, Regionen und Wasserwirtschaft“. Korrelationsanalyse und Varianzanalyse (ANOVA) sind die wichtigsten statistischen Methoden, die auf Zeitreihendaten 2006–2023 basieren. Die Ergebnisse zeigen, dass die jährlichen Schwankungen des Schadholzanfalls wesentlich volatiler sind als jene der Sortimentsangebote, deren Anteile am Einschlag nur mäßig von der Menge an Schadholz beeinflusst werden. Sowohl bei Nadel- als auch bei Laubholz nimmt der Anteil des Sägerundholzes nicht signifikant ab und der Anteil des Faserholzes sowie des Energieholzes nicht signifikant zu, wenn Sturmschäden zunehmen, eher umgekehrt. Wenn allerdings der Schadholzanfall aus anderen Ursachen als Sturm (z. B. Borkenkäfer, Schneebruch) zunimmt, geht der Anteil des Sägerundholzes tendenziell zurück. Bei Nadelholz wirken sich Schäden aus anderen Ursachen als Sturm hauptsächlich und (teilweise) signifikant positiv auf das Angebot an Energieholz aus, während bei Laubholz die wichtigsten und (teilweise) signifikant positiven Auswirkungen auf das Angebot an Faserholz zu verzeichnen sind. Der Anteil von Schadholz liegt beim Endnutzungseinschlag über dem Durchschnitt (größere Dimensionen), beim Vornutzungseinschlag (geringere Dimensionen) unter dem Durchschnitt. Die Korrelationsanalysen zeigen, dass sich zunehmende Kalamitäten stärker auf die Endnutzung als auf die Vornutzung auswirken, so dass das Angebot an Sägerundholz tendenziell leicht zunimmt. Obwohl die Holzeinschlagsdaten aufgrund der Differenzierung nach Eigentumsarten, Sortimenten, Nadel- und Laubholz sowie Schadensursachen (z. B. Stürme, Borkenkäfer) eine detaillierte Analyse erlauben, muss die Interpretation der Ergebnisse eingeschränkt bleiben, da die Datenlage keine Einbeziehung von Sortimentsqualitäten und Sortimentspreisen zulässt. Zudem ist die Länge der Zeitreihe (2006–2023) recht kurz. Es werden mehrere Vorschläge für mögliche weiterführende Forschungsaktivitäten unterbreitet. 

1 Introduction

Forests in Austria are subject to an increasing frequency and magnitude of natural disturbances (Schelhaas et al. 2003) which in general have a significant negative impact on the entire forest-based sector as well as wider socio-economic implications (Gardiner et al. 2013; Donis et al. 2020). The current composition of Austria’s secondary forests exacerbate effects of larger-scale impacts of climate change and already necessitate to induce adaptive forest management measures (Ledermann et al. 2022).

After calamities, which can be broadly distinguished into abiotic damage such as storm and wind breaks and biotic damage such as bark beetle infestations, forest owners usually induce post-disturbance salvage or sanitary logging measures by removing fallen, damaged or infested trees to reduce economic losses (Čakša et al. 2020) and also to comply with forest legislation to prevent further spreading of infections (BMLFUW 2003).

In the period 2006 to 2023 39% of the Austrian timber harvests consisted of salvage logging due to biotic as well as abiotic reasons (BMLFUW 2001-2017; BMNT 2018-2019; BMLRT 2020-2022, BML 2023-2024). Figure 1 shows that total harvests and their slight increase since 2000 have largely been driven by the amount of timber from salvage logging (average annual increase of salvage logging 150.000 cum under bark). Regular harvests – defined here as the difference between total harvests and salvage logging of damaged timber – have been stagnating (average annual decrease 20.000 cum u.b.), which is partly a reaction to compensate for potential timber over-supply and partly to keep harvests within or below a sustainable level. According to the Austrian Forest Inventory 2016-2021 larger forest enterprises already have harvested slightly above net-annual-increment (BFW 2025). In other words, the amount of damage-caused harvests plays an important role regarding the domestic supply of timber in Austria.

Further differentiation of data regarding timber supply from salvage logging (2006-2023) shows that timber from damage caused by windthrows and windbreaks as the main abiotic cause have contributed 42% to the total amount of salvage logging, while bark beetles as the main biotic cause contributed 37% (figure 2).

A bulk of literature exists about reasons for ecological as well as climatic impacts (e.g. carbon balance) of calamities (e.g. Albrecht et al. 2010, Gardiner et al. 2013, Seidl et al. 2014, Thürig et al. 2013), including (silvicultural) management options (e.g. Mason & Valinger 2013, Zimmermann et al. 2018). Other authors focused on the question of risk management, planning support (e.g. Härtl et al. 2013, Holthausen 2004, Holthausen et al. 2004) and risk perception (e.g. Andersson & Gong 2010) in relation to forest damage. Some authors have investigated the possibilities of forest damage insurances (e.g. Sacchelli et al. 2018, Sauter et al. 2016). Baur et al. (2003) as well as Hanewinkel and Peyron (2013) have addressed the economic impact of storms for forest owners either regarding loss of income (loss in wood quality) and/or of the value of the forest estate.

Several authors have used econometrics to estimate the negative impact (elasticities) of a short- or medium/long-term oversupply of timber through calamities on timber prices (Bergen et al. 2002, Mantau 1987, Prestermon & Holmes 2000 and 2004, Schwarzbauer, 2006), which can also be a main reason for the income loss of forest owners. Schwarzbauer and Rauch (2013) added the aspect of a future procurement risk for the forest-based industry due to a medium to long-term reduction of growing stock and increment. The results of the research project „CareforParis“ also show that in the so-called „disturbance scenario“ with an assumed increase of damage caused timber growing stock and increment will decrease quite dramatically in the long run (Weiss et al. 2020, Ledermann et al. 2022).

No study so far exists that deals with the impact of timber supply from salvage logging on the supply of timber assortments, in particular on a potential change in the share of these assortments within total harvests on top of a potential increase of these harvests. Only the deterioration of wood quality is sometimes mentioned (Thürig et al. 2013, Hanewinkel & Peyron 2013). Härtl et al. (2013) use grading options in relation to calamities and price fluctuations as assumptions for their YAFO model, but they are an input, not an output. 

This study aims to analyse whether and how the amount of timber from salvage logging has an impact on the shares of assortments of total (domestic) timber supply as well as on the impact of salvage logging on the relationship between thinnings and final cuts in Austria.

The following research questions and hypotheses are addressed:

1. Does the amount of timber from salvage logging have an impact on the conversion of timber into the assortments of sawlogs, pulpwood and wood for energy? Does this impact differ by ownership categories and wood species group (coniferous vs. non-coniferous)?

Hypothesis 1: An increase in the total amount of timber from salvage logging leads to lower wood quality and therefore to a higher use of wood for energy as well as a lower material use (sawlogs, pulpwood).

2. Do causes of calamities (especially storm vs. other than storm) have an impact on the conversion of timber into intermediate product assortments?

Hypothesis 2: Storms tend to affect more larger trees, snow-break and bark beetles tend to affect more smaller trees. Therefore, an increase in windthrows and windbreaks will create larger timber dimensions and – despite possible lower wood quality – increase the supply of sawlogs rather than wood for energy and pulpwood. An increase in timber supply by other reasons than storms will create smaller timber dimensions and rather increase the supply of wood for energy and pulpwood.

3. Does the amount of timber from salvage logging have an impact on the shares of thinnings and final-cuts in total harvests (as well as vice versa) and does this impact differ by ownership category and wood species group?

Hypothesis 3: Because in terms of timber supply quantity older stands are generally more affected by forest damage than younger stands, an increase in the amount of damage caused timber leads to a shift towards more final-cuts and less thinnings.

2 Methods and Data

Methods

Correlation analysis and analysis of variance (ANOVA) are the main statistical methods used, based on the time-series data 2006-2023 (18 observations). Both focus on the relationship between the shares of total timber from salvage logging in total harvests on the one hand and the shares of assortments as well as thinnings and final-cuts on the other hand. In order to avoid spurious correlation, shares of timber supply from salvage logging as well as assortments with regard to harvests rather than absolute numbers were used in the analyses. In case of parallel or opposing trends of time-series data spurious correlations may appear, which are simply based on trends and not on causal relationships. By using shares rather than absolute numbers the impact of trends as a main reason for spurious correlation can be reduced or even eliminated (see Granger & Newbold 1974). To illustrate the calculation of shares, three examples: Total supply from salvage logging was set in relation to total harvests (all forests, total of coniferous and non-coniferous); coniferous sawlog harvests in SFH were set in relation to total coniferous harvests in SFH, non-coniferous wood for energy harvests in FE were set in relation to total non-coniferous harvests in FE; all other shares in analogy.

ANOVA (t-test with independent samples) was mainly used as a validity-check for the correlation results. The independent variable, the share of total timber from salvage logging (time series from 2006-2023), was recoded into two (ordinal) groups: low = below and equal to, high = above the respective median of the timber from salvage logging share. Both groups include 9 observations. t-tests were used to check whether the means of the dependent variables (share of assortments, share of thinnings and final-cuts) differ significantly between the two groups of low and high share of timber from salvage logging. In addition to the statistical analyses expert interviews with practitioners and specialists in timber markets for the three ownership categories (Friedl 2021 [SFH], Holzer 2021 [FF], Montecuccoli 2021 [FE]), were conducted to further validate and check the plausibility of the results derived from the statistical analysis, but also to get a better background understanding. 

Data

The main data source for this study are the annual „Timber Felling Reports“ (TFRs; „Holzeinschlagsmeldungen“ in German) of the Austrian Federal Ministry of Agriculture, Regions and Tourism (BMLRT, BMLFUW or BMNT, BMLRT, BMLFRW, BML in earlier years and Federal Ministry Agriculture and Forestry, Climate and Environmental Protection, Regions and Watermanagement“ [BMLFKURW) since 2025). Due to a structural break in the data regarding wood for energy, only annual data since 2006 are used in this study for consistency reasons (BMLFUW 2001-2017, BMNT 2018-2019, BMLRT 2020-2022, BML 2023-2024). Before 2006 only one category of wood for energy was reported (fuelwood stacked - „Scheitholz“) in the annual TFRs, which did not include wood chips directly from forests („Waldhackgut“). Only since 2006 both categories are reported separately, although a part of the wood chips have most likely been included in the category fuelwood stacked already before 2006 (see Braun & Schwarzbauer 2018). Due to this inconsistency/structural break only the data since 2006 were used here. All data used for the analysis are presented as supplementary material in tables S1-S4.

TFR data for all harvesting components allow a differentiation by the three ownership categories (small  forest holdings < 200 ha [SFH; „Kleinwald“; 57% of total forest area available for wood supply (FAWS – „Ertragswald“), 59% of growing stock]; forest enterprises > 200 ha [FE; „Betriebe“; 30% of FAWS, 28% of growing stock]; Federal Forests [FF; Österreichische Bundesforste; 13% of FAWS and 13% of growing stock]; numbers according to the Austrian Forest Inventory 2016-2021, BFW 2025) as well as a differentiation by coniferous and non-coniferous timber and a differentiation by assortments (sawlogs, pulpwood and wood for energy; wood for energy is further distinguished between fuelwood stacked [„Scheitholz“] and wood chips from forests [„Waldhackgut“]). The TFRs further distinguish final-cuts and thinnings. In addition, timber supply from salvage logging can be further disaggregated by the reasons for the damage (see figure 2). For the ANOVA-analysis timber supply by salvage logging was grouped into two groups of reasons for the damage: „only storm“ and „other than storm“ (bark beetles as well as other biotic and abiotic causes). Although non-coniferous timber makes up only about 16% of total timber harvests in Austria (Ø 2006-2023; BMLFUW 2001-2017, BMNT 2017-2018, BMLRT 2019-2022, BML 2023-2024) it is important to distinguish coniferous and non-coniferous species in our analysis for several reasons:

• Coniferous and non-coniferous timber are subject to a different demand structure. About 80% of coniferous timber is used for materials (sawnwood, pulp, panels), only about 20% for energy; on the other hand, only about 32% of non-coniferous timber is used for materials, about 68% for energy (BML 2023-2024).
• Coniferous and non-coniferous timber have different technical properties, which is eg. important for the use of timber in construction, in which – according to the current scientific status - coniferous wood cannot yet fully be substituted by non-coniferous wood (see e.g. Espinoza & Buehlmann 2018, Schier et al. 2018).
• Between the forest inventory periods 1992-1996 and 2016-2021 the share of coniferous species in the total stocked forest area available for wood supply has decreased by 5%-points, the share of non-coniferous species has increased by 5%-points accordingly (BFW 2025). Due to climate change this trend is likely to continue and may create challenges for the forest-based industries (in particular sawmills), which are currently mainly processing coniferous timber (Weiss et al. 2020).

The research questions were analysed by using the deepest possible differentiation (by ownership categories, by species groups, by assortments, by final-cuts and thinnings as well as by reasons for the damage).

3 Results

3.1 Relationships between shares of timber from salvage logging and of assortments in harvests 

Figure 3 depicts the annual shares of total timber from salvage logging and the shares of the three assortments (sawlogs, pulpwood round & split and wood for energy) in total timber harvests. It can be seen that the annual fluctuations of timber from salvage logging are much more volatile than those of the assortments. This is a first indication that the shares of the assortments may be only slightly affected by salvage logging.

Figure 4 shows the scatterplots of the shares of total timber from salvage logging and the shares of the assortments in total harvests, each including a best fit straight line and the correlation coefficient. Overall (all ownership categories and total of coniferous and non-coniferous timber), there is no statistically significant correlation between the shares of total timber from salvage logging and the shares of assortments. However – even though not statistically significant – the graph shows that the share of sawlogs tends to increase with an increasing share of timber from salvage logging while the shares of pulpwood and wood for energy tend to decrease. This indicates a contradiction to hypothesis 1 and will be further analysed on a more detailed level.

Table 1 shows the correlations between the share of total timber from salvage logging and the share of assortments in harvests, differentiated by ownership categories as well as by coniferous/non-coniferous wood and by reasons of damage (only storm/other than storm).

Despite some differences regarding the significance of correlation coefficients there are quite consistent - partly surprising – patterns regarding the relationships between the share of total timber from salvage logging and the share of assortments. Differences are not primarily between ownership categories but mainly between coniferous and non-coniferous wood as well as between the reasons for damage (storm vs. reasons other than storm). For all forests there are no statistically significant correlations for coniferous wood between the share of total damage caused timber and the share of the respective assortments. When further broken down into ownership categories and the reasons for damage, there is a clear difference between „only storm“ and „other than storm“. The share of damage caused coniferous timber other than by storm (smaller dimensions) in most cases correlates significantly negative with the supply share of sawlogs and significantly positive with the supply share of wood for energy.

For non-coniferous wood the share of total damage caused timber correlates statistically positive with pulpwood supply and negative with wood for energy supply (exception FF; here the correlation is negative for both, pulpwood and wood for energy, but not significantly). This situation can partly be explained by the fact that the costs for producing wood for energy in the forest is higher than for producing pulpwood. In addition, a decrease in non-coniferous wood for energy can also be caused simply by the fact that more coniferous wood for energy is available. With the exception of the FF a significant decrease of the share of non-coniferous wood for energy corresponds with a significant increase of the share of coniferous wood for energy caused by other reasons than storm. A high amount of coniferous wood for energy, which mainly can be contributed to low quality and cannot be avoided, reduces the willingness of forest owners to produce non-coniferous wood for energy; pulpwood round and split is more attractive (Friedl 2021). There is a clear difference on bucking when the damage is broken down by damage reasons. An increase in the share of storm damage also increases the supply share of non-coniferous sawlogs and decreases the share of non-coniferous wood for energy (partly significantly), an increase in the share of damage caused by other reasons than storms decreases the share of non-coniferous sawlogs (significantly only for all forests and SFH < 200 ha), increases the supply share of non-coniferous pulpwood (FF not significant); no significant relationship with the share of non-coniferous wood for energy. There is also an opposing impact on the two components of non-coniferous wood for energy: other than by storm damaged wood correlates significantly negative with the supply share of wood for energy (stacked), while significantly positive with the supply share of wood for energy chips from the forest (FF not significant).

For both, coniferous as well as non-coniferous wood, the share of sawlogs is not significantly decreasing and the share of pulpwood and wood for energy is not significantly increasing when storm damage is increasing; rather it is the other way around (but rarely significantly).  Due to the higher contribution margin of sawlogs compared to pulpwood and wood for energy, forest owners generally tend to produce as much sawlogs as possible (Friedl 2021, Montecuccoli 2021) and deliver sawlogs earlier than pulpwood and wood for energy (Holzer 2021). When salvage logging in FE exceeds 20% of total harvests, forest owners tend to reduce thinnings in order to meet the allowable cut („Hiebsatz“); this mainly results in a reduction of pulpwood production (Montecuccoli 2021). A further reason for a lower production of pulpwood in times of salvage logging – especially when damages are caused by storm – is the increase of higher sawnwood production of sawmills, which leads to a higher production of sawmill residues and subsequently a lower demand for pulpwood round and split (Friedl 2021, Holzer 2021, Montecuccoli 2021). When the damage caused by other reasons than by storm is increasing, then there generally is a decrease in the share of sawlogs. For coniferous wood the impact of damage from other than by storm is mainly and (partly) significantly positive on the supply of wood for energy, while for non-coniferous wood the main and (partly) significantly positive impact is on pulpwood supply (exception FF). One of the main reasons for the latter may be the fact that non-coniferous pulpwood is more important than coniferous pulpwood for the kraft process in the pulp-industry.

To validate the results of the correlation analysis an analysis of variance (ANOVA) was carried out in addition (see chapter 2). Table 2 shows that different means of the shares of assortments generally are consistent with the correlation coefficients in table 1: a higher mean in the „high“ (share of total timber from salvage logging) category in table 2 in most cases corresponds with a positive correlation coefficient, a lower mean in the „high“ category with a negative correlation coefficient in table 1 and vice versa (the level of significance is not always the same in both tables). In general, the means of assortment shares between the two categories („low“/“high“) do not differ very much in size, in most cases not even significant, which is not really a surprise (see figure 3). „High“ and „low“ shares of coniferous timber from salvage logging do not significantly affect the shares of sawlogs, pulpwood and wood for energy in all ownership categories, even when the shares of timber from salvage logging is further broken down into the „storm“ and other than „storm“ categories. However, the means of coniferous sawlogs in the „high“ category are slightly and insignificantly higher than in the „low“ category for FE > 200 ha and FF. 

More statistically significant differences exist for the means of non-coniferous timber, which is consistent with the correlation results (table 1). Across ownership categories the means of shares for pulpwood are generally slightly higher in the „high“ category and lower in the „low“ category, while it is the other way round for wood for energy. This again confirms the correlation results (table 1) that for non-coniferous timber an increase in timber from salvage logging has a positive impact on the supply of pulpwood and a negative impact on the supply of wood for energy. Only when broken down into the reasons for damage a moderate (statistically insignificant) impact on the supply of sawlogs can be detected. A high non-coniferous share of timber from storm caused salvage logging tends to increase whereas a low share tends to decrease the share of non-coniferous sawlogs. A high share of timber from non-storm caused salvage logging tends to increase, a low share tends to decrease the share of non-coniferous pulpwood. 

3.2 Relationships between the share of total timber from salvage logging and the share of final cuts as well as thinnings 

One explanation that the increasing share of timber from salvage logging is not decreasing, but in most cases (non-significantly) increasing the supply share of sawlogs and decreasing the supply share of pulpwood and/or wood for energy lies in the fact that the supply of timber from salvage logging is above the average in final cuts and below average in thinnings. Therefore, the share of sawlogs (in particular for coniferous timber) and pulpwood (in particular for non-coniferous timber) can positively correlate with the share of timber from salvage logging. These relationships can also be validated by various assortment tables and bucking analyses (see e.g. Sterba & Grieß, 1983; Eckmüllner et al. 2007) and have been confirmed by interviews (Friedl 2021, Holzer 2021, Montecucoli 2021).

Table 3 shows the correlation results between the shares of total timber from salvage logging and final cuts, further disaggregated by coniferous and non-coniferous timber. With the exception of the FF, where comparatively low correlations exist and only for non-coniferous timber, there are highly significant and strong correlations. Although we have no empirical prove, we suspect that the low level of correlations in FF can partly be explained by the fact that the FF is a forest enterprise structured into ten regional enterprises distributed over most of Austria. To meet the economic interests of the entire company regional enterprises, which are less affected by a calamity, could reduce their harvests or decrease their final cuts and increase thinnings and compensate for salvage logging in more affected enterprises. But in general, the correlations show that calamities affect final cuts more than thinnings.

The ANOVA results (table 4) confirm the results of the correlation analysis with the exception that there is no significant difference between low and high share of timber from salvage logging in FF for both, coniferous and non-coniferous timber. The levels of means for final cuts are all (coniferous, non-coniferous and total) significantly different for SFH < 200 ha, FE > 200 ha and all forests.

4 Discussion and Conclusions

This section includes answers to the research questions and hypotheses, a critical assessment of the research results as well as some follow-up options for further research.

Answering the research questions and research questions not asked

Research question/hypothesis 1: Does the amount of total timber from salvage logging have an impact on the conversion of timber into the assortments of sawlogs, pulpwood and wood for energy? Does this impact differ by ownership categories and wood species groups (coniferous – non-coniferous)?

There are no statistically significant correlations between the share of total coniferous timber from salvage logging and the share of the respective assortments for all ownership categories. Regarding coniferous timber hypothesis 1 has to be rejected.

For non-coniferous timber and by ownership categories the share of total timber from salvage logging correlates statistically positive with pulpwood supply (not for FF) and negative with wood for energy supply; except for FF there is no significant correlation with sawlog supply. The positive correlation with pulpwood supply and the negative correlation with wood for energy can be (partly) explained with higher costs for fuelwood procurement and also the importance of non-coniferous pulpwood for the pulp-industry (kraft process). Regarding non-coniferous timber hypothesis 1 has to be rejected regarding the supply of sawlogs and wood for energy but not for the supply of pulpwood.

Research question/hypothesis 2: Do causes of calamities (especially storm vs. other than storm) have an impact on the conversion of timber into intermediate product assortments?

There is a clear difference between the causes „only storm“ and „other than storm“. Although in most cases not statistically significant, storm damaged coniferous timber (larger dimensions) correlates positively with the sawlog supply share and negatively with the shares of pulpwood and wood for energy. The share of coniferous timber from salvage logging other than by storm (smaller dimensions) correlates (partly) significantly negative with the supply share of sawlogs and (partly) significantly positive with the supply share of pulpwood and/or wood for energy. 

An increase in the share of non-coniferous storm damage also increases the supply share of non-coniferous sawlogs and decreases the share of  non-coniferous wood for energy (partly significantly; exception FF), an increase in the share of damage caused by other reasons than storms decreases the share of non-coniferous sawlogs (partly significant), and significantly increases the supply share of non-coniferous pulpwood (positive, but insignificant for FF); no significant relationship with the share of non-coniferous wood for energy. 

Hypothesis 2 can therefore not be rejected for both coniferous and non-coniferous timber.

Research question/hypothesis 3: Does the amount of total timber from salvage logging have an impact on the shares of thinnings and final-cuts in total harvests (as well as vice-versa) and does this impact differ by ownership categories and wood species group? 

With the exception of FF, where comparatively low correlations exist and only for non-coniferous timber, there are highly significant and strong correlations. These correlations indicate that calamities affect older stands – and therefore – final cuts more than younger stands and thinnings. This is also an explanation that in some categories increased damage caused timber is also (surprisingly and mostly insignificantly, however) leading to a higher supply of sawlogs. 

Hypothesis 3 can therefore not be rejected.

Originally, we had two more research questions: Are there time-lags between the occurrence of calamities and the emergence of timber from salvage logging, by causes of calamities and/or by assortments? Our hypothesis was that the damage by bark beetles many times occurs in the aftermath and as a lagged result of storm damage (see e.g. Suliman & Ledermann 2025), which may lead to a lag in the bucking of pulpwood and wood for energy, following the immediate bucking of an increased amount of sawlogs. However, the data situation is a barrier to address this research question. The amount of salvage logging is reported in the TFRs in the same year as all other harvests. However, if or how much salvage timber is not reported in the same year but carried over to the next year, is unknown. However, we did carry out correlation analyses of lagged salvage timber amount with the distribution of assortments in the respective following years; no statistically significant correlations could be established. We also analyzed whether the amount of salvage timber caused by storms/wind-breaks in the current years is correlated to the amount of salvage logging by other reasons (bark beetle) in the respective following year. No statistical significance could be established.

Another research question would have been to check, whether and to what extend the amount of assortments from salvage logging (e.g. sawlogs from salvage logging) has an impact on the regular harvest of assortments (e.g. on the amount of sawlogs from regular harvests). This research question could not be addressed, because the data for salvage logging do not distinguish between assortments – they only include the total of the assortments in the respective categories.

Critical assessment of the results

This analysis is mainly based on data of the Austrian Timber Felling Reports (TFRs), the data quality of which is not undisputable, in particular for the ownership category SFH < 200 ha (see e.g. Ettwein et al. 2015). But this is the only existing database which offers annual time series with such a high level of differentiation regarding ownership categories, species, assortments, final cuts/thinnings, timber from salvage logging (incl. reasons for damage). Considering the limits, it is therefore the only database for such an analysis in order to avoid time-consuming and very expensive primary data surveys.

In addition, the TFRs do not report logging waste that remains in the forest. It can be assumed that the amount of logging waste is higher in cases of calamities and salvage logging compared to regular harvests. This could contribute to a higher share of sawlogs and a lower share of wood for energy in salvage loggings.

Another problematic aspect of this analysis is the length of the time series. Due to inconsistencies in the data (structural break, see chapter 3.) we only could use 18 annual observations. On the other hand, the share of timber from salvage logging in total harvests between 2006 and 2023 is much higher (39%; BMLFUW 2001-2017, BMNT 2017-2018, BMLRT 2019-2022, BML 2023-2024) than in the period 1996-2005 (29%; BMLFUW 2001-2017), which makes the analysis quite focused.

There are two more shortcomings in the data. The categories of the assortments in the TFRs are not reported considering quality and prices. Potential lower wood quality and prices are not at all reflected in the data, which would be especially relevant for the assortment sawlogs (e.g. increased amount of sawlogs in (low) Cx quality). Furthermore, damage caused storm timber cannot be further disaggregated into windthrows and windbreaks, which also is an important criterion for the quality, again in particular for sawlogs. According to Friedl (2021) and Holzer (2021) windbreaks lead to a higher share of wood for energy, while windthrows lead to higher share of sawlogs. Windbreaks usually happen in the lower parts of the stems and therefore reduce the potential for sawlog production, while windbreaks usually do not affect wood quality and therefore do not necessarily reduce the potential for sawlog production (Montecuccoli 2021).

Potential follow-up research activities

Due to the existing limitations regarding this analysis several follow-up research activities can be considered:

The analysis could be repeated at a later stage with longer time series and additional qualitative expert interviews with foresters and sales persons in the roundwood market could be conducted to better complement the statistical analysis of secondary data. Primary data surveys on specific bucking activities on damaged forest areas could be conducted, in particular considering the aspect of windthrows vs. windbreaks, but also the aspect of calamities affecting larger areas vs. calamities affecting single trees or smaller pockets of trees. In addition, the aspects of quality and prices of assortments, relevant for bucking decisions, could be addressed. For FE a specific analysis of the Austrian Forest Accountancy Data Network [Testbetriebsnetz] (see e.g. Sekot & Metzker 2024) could be revealing, because these data include – among many other aspects – also the share of salvage logging in timber harvests, broken down by thinnings and final cuts and – at least for some FE - data on the share of sawlogs in Cx quality (low sawlog quality). For FF the internal (unpublished) database could be analysed – if made available -, because these most likely include in a consistent way and for a long time period data on storage of salvage logged timber, and also the share of low quality sawlogs. Not a direct research option but a data improvement possibility: The TFRs could be complemented with data disaggregating the storm damage data by windthrows and windbreaks as well as data regarding calamities on larger areas vs. calamities affecting single trees or smaller pockets of trees (see above). In addition, the TFRs could distinguish salvage logging amounts into assortments. However, in both cases the validity of these data may be questionable.

Acknowledgements

This research was carried out without specific external funding and is entirely based on the personal research interests of the authors. The authors thank Marilene Fuhrmann (BEST, Sustainable Supply and Value Cycles) as well as Raphael Asada (University of Graz, Institute of Systems Sciences, Innovation, and Sustainability Research – SIS) for their valuable comments regarding the manuscript. We also thank the three interview partners for their valuable input: Wolfgang Holzer (FF – Österreichische Bundesforste), Klaus Friedl (SFH – LK Steiermark), Felix Montecuccoli (FE – Land- und Forstbetriebe Österreich). Last, but not least we thank the two reviewers for their critical comments.

References

Albrecht A., Hanewinkel M., Bauhus J., Kohnle U., Nabuurs G-J., Rosén K., Lindner M., Paschalis-Jakubowicz P., Mayer H., Schindler D., 2010. How does silviculture affect storm damage in forests of south-western Germany? Results from empirical modeling based on long-term observations. Berlin/Heidelberg: Springer Science and Business Media LLC. European Journal of Forest Research, 2010-09-08, Vol.131 (1), pp 229-247

Andersson M., Gong P., 2010. Risk preferences, risk perceptions and timber harvest decisions — An empirical study of nonindustrial private forest owners in northern Sweden. Forest Policy and Economics, Volume 12, Issue 5, June 2010, pp 330-339

Baur P., Holthausen Chr.-N., Roschewitz A., 2003. LOTHAR. Ökonomische Auswirkungen des Sturms Lothar im Schweizer Wald, Teil II. Verteilung der Auswirkungen auf bäuerliche und öffentliche WaldeigentümerInnen: Ergebnisse einer Befragung [Economic Impacts of Storm Lothar in Swiss Forests, Part II: Allocation of Impacts on Small Private and Public Forest Owners; Results of a Survey]. Umwelt-Materialien Nr. 158. Bundesamt für Umwelt, Wald und Landschaft, Bern. 

Bergen V., Löwenstein W., Olschewski R., 2002. Forstökonomie. Volkswirtschaftliche Grundlagen [Forest Economy. Macroeconomic Basics]. Verlag Franz Vahlen, München.

BFW – Bundesforschungszentrum Wald, 2025. Österreichische Waldinventur, Erhebungen 1992-1996 und 2016-2021 [Austrian National Forest Inventory, periods 1992-1996 and 2026-2021].  https://www.waldinventur.at/?x=1486825&y=6059660&z=6.86909&r=0&l=1111#/map/2/tp09_1621_XXX_001_1/Bundesland/erg9 (access date: January 10th, 2025)

BMLFUW - Bundesministerium für Land- und Forstwirtschaft, Umwelt und Wasser, 2001-2017. Holzeinschlagsmeldung(en) [Timber Felling Reports] 2000-2016, Eigenverlag, Wien

BMLFUW - Bundesministerium für Land- und Forstwirtschaft, Umwelt und Wasser, 2003. Verordnung des Bundesministers für Land- und Forstwirtschaft, Umwelt und Wasserwirtschaft über den Schutz des Waldes vor Forstschädlingen (Forstschutzverordnung) [Decree of Federal Minister of Agriculture, Environment and Water Management Regarding the Protection of Forests from Forest Pests]. StF: BGBl. II Nr. 19/2003, Wien

BML - Bundesministerium für Landwirtschaft, Regionen und Wasserwirtschaft, 2023-2024. Holzeinschlagsmeldung(en) [Timber Felling Reports] 2022-2023, Eigenverlag, Wien

BMLRT - Bundesministerium für Landwirtschaft, Regionen und Tourismus, 2020-2022. Holzeinschlagsmeldung(en) [Timber Felling Reports] 2019-2021, Eigenverlag, Wien

BMNT - Bundesministerium für Nachhaltigkeit und Tourismus, 2018-2019. Holzeinschlagsmeldung(en) [Timber Felling Reports] 2017-2018, Eigenverlag, Wien

Braun M., Schwarzbauer P., 2018. Eigenverbrauch und Marktverfügbarkeit von Energieholz aus dem Wald [Subsistence Use and Market Availabilty of Energy Wood from Forests]. Unpublished project report commissoned by Kooperationsplattform Forst-Holz-Papier (FHP), Working Group Energy, Vienna

Čakša L., Šēnhofa S., Šņepsts G., Elferts D., Liepa L., Jansons Ā., 2020. Effect of Stem Snapping on Aspen Timber Assortment Recovery in Hemiboreal Forests. Forests 12, 1-28

Donis J., Saleniece R., Krisans O., Dubrovskis E., Kitenberga M., Jansons A., 2020. A Financial Assessment of Windstorm Risks for Scots Pine Stands in Hemiboreal Forests. Forests 11(5), 1-10

Eckmüllner O., Schedl P., Sterba H., 2007. Neue Schaftkurven für die Hauptbaumarten Österreichs und deren Ausformung in marktkonforme Sortimente [New Shaft Curves for the Main Tree Species in Austria and their Transformation into Market Conform Assortments]. Centralblatt für das gesamte Forstwesen. Wien 124(3-4): 215-236

EspinozaO., Buehlmann U., 2018. Cross-Laminated Timber in the USA: Opportunity for Hardwoods? Current Forestry Reports (2018) 4: 1–12. https://doi.org/10.1007/s40725-018-0071-x

Ettwein F., Schwarzbauer P., Hesser F., Stern T., 2015. Vertiefte Datenanalyse. Projektendbericht für FHP [Deepend Data Analysis. (Wood K plus Projektnummer NK 14 004). Unpublished project report commissoned by Kooperationsplattform Forst-Holz-Papier (FHP), Wien

Friedl K., 2021. Personal digital communication via Zoom software in his capacity as the responsible officer for forests and wood markets within the Chamber of Agriculture Styria. Date: March 1st, 2021, Vienna and Graz

Gardiner B., Schuck A., Schelhaas M-J., Orazio Chr., Blennow K., Nicoll B. (editors), 2013. Living with Storm. Damage to Forests. European Forest Institute. What Science Can Tell Us 3, 129

Granger C.W.J, Newbold P., 1974. Spurious regressions in econometrics. In: Journal of Econometrics. Nr. 2, 1974, pp 111–120,

Hanewinkel M., Peyron J-L., 2013. The economic impact of storms. In: Gardiner et al., 2013. Chapter 3.2., pp 55-63

Härtl F., Hahn A., Knoke Th., 2013. Risk-sensitive planning support for forest enterprises: The YAFO model. Computers and Electronics in Agriculture 94, pp 58–70

Holthausen N., 2004. Ökonomische Bedeutung und Management von Naturrisiken im Wald: Theoretische Grundlagen und empirische Analysen nach dem Sturm Lothar in der Schweiz [Economic Impact and Management of Natural Risks in Forests. Theoretical Basics and Empirical Analyses after Storm Lothar in Switzerland]. Inaugural-Dissertation, Albert-Ludwigs-Universität, Freiburg i.Br.

Holthausen N., Hanewinkel M., Holecy J., 2004. Risikomanagement in der Forstwirtschaft am Beispiel des Sturmrisikos [Management of Risks in Forestry – the Example of Risk by Storm]. Forstarchiv 75 (2004), pp 149–157

Holzer W., 2021. Personal digital communication via Team software in his capacity as the leading manager for timber marketing of the Austrian Federal Forests. Date: February 3rd, 2021, Vienna and Purkersdorf

Ledermann T., Kindermann G., Jandl R., Schadauer K., 2022. Klimawandelanpassungsmaßnahmen im Wald und deren Einfluss auf die CO2-Bilanz [Measures to Mitigate Climate Change in Forests and their Impact on the Carbon Balance]. In: BFW-Praxisinformation Bd. 51, pp 6-13

Mantau U., 1987. Simulation erhöhter Einschlagsmengen infolge neuartiger Waldschäden und ihre ökonomischen Folgen für den Nadelschnittholzmarkt. Forschungsvorhaben „Folgen der neuartigen Waldschäden für die Holzmärkte der Bundesrepublik Deutschland“. Ergebnisbericht 3. [Simulation of Increased Timber Harvests due to New Forest Damages and their Economic Consequences for the Coniferous Lumber Market. Research Project „Consequences of New Forest Damages for the Wood Markets in Germany“. Report 2]. Institut für Forstpolitik und Raumordnung, Arbeitsbereich Holzmarktlehre an der Universität Freiburg, Freiburg i.Br.

Mason B., Valinger E., 2013. Managing forests to reduce storm damage. In: Gardiner et al., 2013. Chapter 4.2., pp 87-95

Montecuccoli F., 2021. Personal digital communication via Zoom software in his capacity as the president of the Association of the Agriculture and Forestry Companies Austria. Date: February 24th, 2021, Vienna

Prestemon J.P., Holmes Th. P., 2000. Timber Price Dynamics Following a Natural Catastrophe. Wiley. American journal of agricultural economics, 2000-02, Vol.82 (1), pp 145-160

Prestemon J.P., Holmes Th. P., 2004.  Dynamics and Optimal Timber Salvage After a Natural Catastrophe. Bethesda: Society of American Foresters. Forest Science, 2004-08-01, Vol.50 (4), p 495

Sacchelli S., Cipollaro M., Fabbrizzi S., 2018. A GIS-based model for multiscale forest insurance analysis: The Italian case study. Forest Policy and Economics, Volume 92, July 2018, pp 106-118

Sauter Ph.A., Möllmann T.B., Anastassiadis F., Mußhoff O., Möhring B., 2016.  To insure or not to insure? Analysis of foresters' willingness-to-pay for fire and storm insurance. Forest Policy and Economics, Volume 73, December 2016, pp 78-89

Schelhaas M.-J., Nabuurs G.-J., Schuck A., 2003. Natural disturbances in the European forests in the 19th and 20th centuries: NATURAL DISTURBANCES IN THE EUROPEAN FORESTS. Global Change Biology 9, 1620–1633. 

Schier F., Morland Ch., Janzen N., Weimar H., 2018. Impacts of changing coniferous and non‑coniferous wood supply on forest product markets: a German scenario case study. European Journal of Forest Research: 137, pp 279-300

Schwarzbauer P., 2006. Einflüsse von Schadholzmengen auf Rohholzpreise. Eine quantitativ-statistische Analyse am Beispiel Österreichs [Impacts of Damage Caused Timber Harvests on Roundwood Prices. A Quantitative-Statistical Analyses with the Example of Austria]. Allg. Forst- u. J.-Ztg., 178. Jg., 1, pp 2-8.

Schwarzbauer P., Rauch P., 2013. Impact on industry and markets – roundwood prices and procurement risks. In: Gardiner et al., 2013. Chapter 3.3., pp 64-69

Seidl R., Schelhaas M.-J., Rammer W., Verkerk P.J., 2014. Increasing forest disturbances in Europe and their impact on carbon storage. Nature Climate Change. Letters, http://dx.doi.org/10.1038/nclimate2318 (access date: October 12th, 2020)

Sekot W., Metzker M., 2024. Das Testbetriebsnetz im österreichischen Großwald [The Austrian Forest Accountancy Data Network]. AFZ, der Wald: allgemeine Forst-Zeitschrift für Waldwirtschaft und Umweltvorsorge, 79(23):29-32.

Sterba H., Grieß O., 1983. Sortenertragstafeln für Fichte [Assortment Yield Tables for Spruce]. Österreichischer Agrarverlag, Wien

Sulimann T., Ledermann Th., 2025. Modelling salvage cuts in Austrian Forests – I: Wind and snow induced tree breakage. Forest Ecology and Management: 578, https://doi.org/10.1016/j.foreco.2024.122479

Thürig E., Hagedorn F., Lindroth A., 2013. Influence of storm damage on the forest carbon balance. In: Gardiner et al., 2013. Chapter 3.1., pp 47-54

Weiss P., Braun M., Fritz D., Gschwantner T., Hesser F., Jandl R., Kindermann G., Koller T., Ledermann T., Ludvig A., Pölz W., Schadauer K., Schmid B.F., Schmid C., Schwarzbauer P., Weiss G., 2020. Final Project Report „CareforParis“. Climate and Energy Fonds, Vienna

Zimmermann K., Schuetz T., Weimar H., 2018. Analysis and modeling of timber storage accumulation after severe storm events in Germany. Berlin/Heidelberg: Springer Science and Business Media LLC. European journal of forest research, 2018-05-04, Vol.137 (4), pp 463-475

Supplementary Material

Data used for the statistical analysis.