Composition and Elevation of Spruce Forests Affect Susceptibility to Bark Beetle Attacks: Implications for Forest Management

The spruce bark beetle, Ips typographus (L.) (Coleoptera: Curculionidae, Scolytinae), is one of the most destructive insects infesting spruce forests in Europe. Data concerning infestations of I. typographus occurring over the last 19 years (1994–2012) on the Southern Alps were analyzed in seven spruce forest types: (1) pure spruce plantations; (2) pure spruce reforestations; (3) pure spruce mountain forests; (4) pure spruce alpine forests; (5) spruce-conifer mixed forests; (6) spruce-broadleaf mixed forests; and (7) spruce-conifer-broadleaf mixed forests. The collected data included the amount of I. typographus damage and the location and composition of the infested forests. The results indicate that different forest types are differently susceptible to I. typographus. Plantations, reforestations and mountain spruce forests show mean damage and mean number of infestations higher than other forest types. Within pure spruce forests, alpine forests growing at high elevations (>1300 m) suffer low damage. Furthermore, the mean number of infestation spots recorded annually in the different spruce forest types is negatively correlated with a Naturality Index value. The results suggest that forest composition and elevation are the main factors driving the risk of I. typographus damage. A new management strategy for some spruce forest types is needed, with a progressive reduction of pure spruce forests at low altitude and an increase of broadleaf composition.


Introduction
Forest composition and temperature are among the main environmental factors affecting tree suitability to herbivorous insects and, hence, management decisions concerning the control of forest pests [1].Knowledge on the spatio-temporal distribution and population density of the major forest pests should be the first step towards an integrated forest management focused on forest protection and timber production.According to the theory of predisposition and trigger [2], insect outbreaks rely on a spatial and temporal coincidence of susceptible (i.e., predisposed) forests with disturbance factors affecting tree survival and the production of ecosystem goods and services [3].Although foresters cannot determine the occurrence of unpredictable disturbance factors, they can affect the forest health by accurate silvicultural planning and management [4].The poor health conditions of forests growing in sites characterized by environmental traits unsuitable for a particular tree species, such as frequent climatic anomalies, elevation, soil characteristics and forest composition, are expected to trigger insect infestations, with potentially serious consequences on forest survival [5,6].
Bark beetles (Coleoptera: Curculionidae, Scolytinae) are probably the most important pests affecting Holarctic forests [7,8].Many studies on the population dynamics of bark beetles suggest that climatic anomalies, such as drought and heat waves, may trigger large insect outbreaks [8][9][10], acting both directly on the insect populations and indirectly on the health of the potential host trees.In the last few years, the spruce bark beetle, Ips typographus (L.), one of the most destructive insects infesting spruce forests (Picea abies Karsten), caused serious damage almost everywhere in Europe [3,[11][12][13].
Severe damage was also recorded in spruce forests growing on the Southern Alps [14][15][16][17][18][19].Large I. typographus outbreaks were observed mainly in sites with environmental conditions not fully suitable for spruce.Recent studies carried out on the south-eastern Alps show, for instance, a mean damage per hectare about seven times higher in spruce forests growing in sites warmer than those within the natural climatic range of spruce [19].Moreover, the intensity and altitudinal distribution of the insect outbreaks were affected by dry summers, which were related to an increase of the mean damage [17] and an upward shift of the altitudinal outbreak range [19].A preliminary investigation carried out in the same region reported that silvicultural practices may substantially modify the susceptibility of spruce trees to I. typographus infestations [20].Damage caused by I. typographus was recorded in different types of spruce forests, showing a strong variation among the monitored forest types.However, the recorded variables were just some of the main parameters that could determine the availability of weakened or dying host trees and, hence, the type of spruce forests [20]; so, a more detailed analysis is required.
Better knowledge of the relationships between forest composition and pest outbreaks could help foresters in applying silvicultural practices suitable for minimizing the present and future damage due to forest pests.Decision-making in forest pest management calls for tools that can help in understanding the relationships among predisposing factors, forest characteristics and pest occurrence [4], improving the pest control strategies and reducing general costs.In this study, we hypothesize that forest composition, together with climatic conditions, may affect I. typographus-induced damage, and we assume that the damage is lower in high-diversity Norway spruce forests and in stands growing at higher elevations, which is considered a proxy for temperature.The aim of the paper is, therefore, to investigate the differences in the severity of the spruce bark beetle attacks among seven types of spruce forests growing on the Southern Alps and differing mainly in tree species composition, tree origin (i.e., from natural seedlings or planted) and elevation.

Study Species
The spruce bark beetle, I. typographus, is one of the main European forest pests [21,22].The species causes great damage also to spruce stands growing along the Southern Alps [11,[14][15][16]18].Adults of the parental generation leave the overwintering sites in spring (May), when the mean air temperature is about 18 °C [23], and fly looking for suitable hosts, i.e., stressed or dying spruces, where they begin the first generation.In general, I. typographus populations living along the Southern Alps are bivoltine [14][15][16]18].At the beginning of July, the offspring emerges and starts a second generation, which will end in September.After overwintering under the bark or in the litter [24], the offspring of the second generation will emerge in the spring of the following year.
Ips typographus usually infests stressed or dying trees and can respond quickly to the abundance of suitable hosts induced by weather events and unfavorable stand conditions [11].During infestations, I. typographus has a natural tendency to aggregate and produce easily recognizable clusters of infested trees (hereafter called -spots‖).In the following years, spots enlarge, partially affecting trees that have recovered from stress, also.The outbreak dynamics are often driven by negative feedback mechanisms, such as intraspecific competition [11].As I. typographus kills trees to reproduce, tree mortality is positively related to population abundance [14,25].

Study Area and Naturality Index
The study was conducted in the Friuli Venezia Giulia region (NE Italy), a mountainous area of 7844 km 2 located along the southern border of the Alps (Figure 1).Norway spruce is the main forest species in this area, covering about 66,100 ha scattered in 7 main types of mixed and pure forests (Table 1), where spruce is usually growing along the southern edge of its climatic range, as defined by Schmidt-Vogt [26].Del Favero et al. [27,28] calculated a Naturality Index for each of the 7 types of spruce forests.The Naturality Index, calculated following the protocol proposed by Poldini [29] and Poldini and Pertot [30], assesses the natural origin of the forest in relation to the occurrence of human modifications [31,32].In a scale of 1-5, the Naturality Index indicates a status close to (high values) or far from (low values) the natural origin of the forests.Del Favero et al. [27,28] inspected all stands belonging to a specific forest type, assigning a Naturality Index to each stand; all values were then averaged, giving the mean Naturality Index of each forest type [27,28].

Spruce Forest Types
According to the Habitat Classification of the European Nature Information System (EUNIS) [33] and Del Favero et al. [27,28], spruce forests in the Friuli Venezia Giulia region can be classified as follows: Figure 1.Italian region where the study was carried out.For the legend of the forest type codes, see Table 1.
Table 1.Main types of spruce forests occurring in the investigated area and their characteristics.The values of dendrometric parameters are reported as mean ± standard deviation.natural climatic range of spruce [26].In these habitats, spruce often replaces native forests of broadleaves after their felling.They are stands with high tree density (about 350 trees per ha) and without brushwood.This spruce formation covers about 4500 ha, representing ca.6.8% of the spruce forests of the investigated area.They are man-made forests with a very low value (1.0) of the Naturality Index [27,28] (Table 1).
(2) Pure spruce reforestations (REF): natural pure spruce forests growing at about 900 m a.s.l.(range 650-1200), where they replace mixed broadleaf forests or grasslands and meadows.These formations result from past silvicultural practices supporting the spread of spruce by the periodic elimination of native oak, chestnut and beech forests.They cover about 4800 ha, representing ca.7.3% of the total spruce forests of the region; a Naturality Index of 1.3 (Table 1).
(3) Pure spruce mountain forests (MOU): pure and even-aged spruce forests growing on fertile acidophil soils.Natural formations usually growing along steep slopes at about 1000 m a.s.l.(range 700-1300 m), with high tree density and scarce brushwood.Covering only about 3700 ha, MOU are the rarest spruce formations in Friuli Venezia Giulia (less than 6%); a Naturality Index of 2.5 (Table 1).
(4) Pure spruce alpine forests (ALP): natural high-elevation pure spruce forests growing in the most northern areas at about 1500 m a.s.l.(range 1350-1700), where temperatures are generally cold.Here, spruce grows in its best conditions [26].This type of spruce formation covers about 10,500 ha, representing ca.15.7% of all the spruce forests in the region.The low tree density (about 250 trees per ha) and rich brushwood result in a high (4.0)Naturality Index (Table 1).
(5) Spruce-conifer mixed forests (Mix C): natural conifer mixed forests, where spruce grows with either larch, fir or Scots pine, according to local climate and soil conditions.These formations occur at about 1000 m a.s.l.(range 700-1200), with the dominance of different conifer species depending on the habitat conditions and previous forest management.They cover about 7500 ha, representing ca.11.3% of the spruce forests.Large and old trees are common; a Naturality Index of 3.0 (Table 1).
(6) Spruce-broadleaf mixed forests (Mix B): natural spruce formations mixed with broadleaves, usually beech, mostly growing at about 1200 m a.s.l.(range 950-1400) along south-facing slopes, where climatic conditions are favorable to both of the species.The proportion of spruce and broadleaves is affected by past silvicultural practices, local soil conditions and air humidity.They are usually growing on carbonate soils with high water availability, due to frequent rainfall.The natural association between spruce and beech is the most common forest type in the region, covering a surface of about 24,200 ha, i.e., more than a third (36.6%) of the spruce forests; a Naturality Index of 3.8 (Table 1).( 7) Spruce-conifer-broadleaf mixed forests (Mix C-B): natural uneven-aged spruce forests mixed with fir and beech.These are forests growing at about 1200 m a.s.l.(range 1000-1350) on deep and fertile soils, with the proportion of spruce, fir and beech, depending on the local microclimate and silvicultural conditions.This forest type usually grows in sites having silicate soils with reduced water availability and managed with limited silvicultural pressure.They are the second widest spruce formations in Friuli Venezia Giulia, involving about 10,800 ha and representing ca.16.4% of the spruce forests.The occurrence of many tree species in the same site results in a high (4.0)Naturality Index (Table 1).
The mean rainfall regime is similar in all forest types, with about 1200 mm per year.

Data Collection
Data on I. typographus infestations were obtained from the phytopathologic forest inventory of the Friuli Venezia Giulia region (BAUSINVE) [34].Since 1994 BAUSINVE provides information about forest pests and diseases occurring in the regional forests [35].Forest health of the whole region is monitored daily via ground-based survey by about 60 foresters working for the Regional Forest Service and supervised by a team of entomologists and pathologists from scientific institutes.When an outbreak occurs, foresters compile specific reports, including information concerning both the infesting pest (species, development instar, population density) and the total damage (number of trees attacked, volume of infested timber, defoliation area).During the survey, the foresters also classify the characteristics of the attacked forests, collecting data on elevation, tree density, composition, structure, type of forest and management.Data are then recorded and stored in an electronic database.Our database thus derives from a long-term intensive forest health monitoring covering a 19-year period (1994)(1995)(1996)(1997)(1998)(1999)(2000)(2001)(2002)(2003)(2004)(2005)(2006)(2007)(2008)(2009)(2010)(2011)(2012).Each recorded infestation spot composed of at least 5 infested trees was visited to identify the biotic agent causing tree mortality, recording the number of killed trees, the volume of timber loss (m 3 ) and the site characteristics.Using these data, we built a time series of annual timber loss for the whole region by summing the volume (m 3 ) of trees killed in all the spots occurring each year in each forest type.Annual timber loss also included spots detected in the spring of the following year to allow for the delay between late-summer beetle infestations by second or even third generations [18] and canopy discoloration occurring in the next spring.

Statistical Analysis
We tested the hypothesis that all spruce forest types were equally susceptible to I. typographus infestations, by comparing the mean damage (m 3 ) recorded in the last 19 years in the monitored spruce forests.The mean annual volume of timber infested by I. typographus was used as the main dependent variable, because this was carefully measured and is of high practical utility for foresters, but also because it is a reasonable proxy for beetle abundance and, therefore, allows for qualified inferences regarding population dynamics.However, because the compared spruce forest types have different extensions (Table 1), the mean annual volume recorded for each spruce forest type was reported to be a standard area of 1000 ha (m 3 /1000 ha).Because the susceptibility of a forest to a pest may also be assessed considering the temporal infestation frequency and the size of each single infestation event, the different spruce forest types were also compared for the mean number and size of the spots recorded during the 19-year survey.Analysis of variance (ANOVA) was used to find significant differences among spruce forest types in terms of mean annual damage, mean annual number of spots and mean spot size [36].The homogeneity of variance was tested by Cochran's test (test C) and normality by the Kolmogorov-Smirnov test (test D).When necessary, data were log-[X' = log(x + 1)] or arcsin-(X' = arcsinx) transformed to obtain the homogeneity of the variance and normality.Wherever significant differences occurred, Tukey's honestly significant difference (HSD) multiple comparison test was applied for mean separation [36].The mean number of spots reported every 1000 ha was lastly correlated by a linear regression model with the degree of naturality of the monitored types of spruce forests.An R 2 value, adjusted for the number of parameters [36], was used to assess the goodness-of-fit of the tested model.Differences at the 0.05 confidence level were considered significant.Analyses were performed by Statistica for Windows (Statistica, Tulsa, OK, USA).

Results
In the 19-year monitoring, 797 infestation spots of I. typographus were recorded, for a total of 30,475 infested spruce trees (1604 per year) and about 37,780 m 3 (1988 per year) of damaged timber.As hypothesized, different forest types were differently susceptible to I. typographus infestations, suffering damage of different intensity (ANOVA, df = 6; 119, F = 3.84, p < 0.001).In particular, PLA, REF and MOU showed a mean damage significantly higher than the other forest types (Tukey test, p < 0.05), with values close to or greater than 100 m 3 of timber lost annually per 1000 ha (Figure 2).Instead, ALP, Mix B and Mix C-B were less prone to I. typographus infestations, with a mean damage about 10 times lower than the previous types.Finally, Mix C had damage of about 35 m 3 of timber per year every 1000 ha of forest (Figure 2).Different letters indicate significant differences at p < 0.05.For the code legend, see Table 1.
The mean annual number of infestation spots recorded every 1000 ha also significantly differed among forest types (ANOVA, df = 6; 119, F = 4.30, p < 0.001) (Figure 3), showing a very similar trend to those reported for annual damage.PLA, REF and MOU had a mean number of infestation spots higher than other forest types (Tukey test, p < 0.05), with values close to or more than 10 spots per year every 1000 ha (Figure 3).Instead, ALP, Mix B and Mix C-B suffered the lowest numbers of infestations.Again, Mix C showed a value in between (Figure 3).Interestingly, the size of the infestation spots was similar in all forest types (ANOVA, df = 6; 794, F = 0.57, p = 0.75), with a mean value of about 40 m 3 per spot (Figure 4).Different letters indicate significant differences at p < 0.05.For the code legend, see Table 1.For the code legend, see Table 1.
The mean number of infestation spots recorded annually in the different spruce forest types was negatively correlated with the Naturality Index of the monitored forests (F = 25.56;R 2 = 0.83; p < 0.01) (Figure 5), with spruce forests intensively managed, such as PLA and REF (Table 1), suffering the highest number of I. typographus spots.

Discussion
We found that different forest types suffer damage of different severity.Spruce forests, usually characterized by low diversity (a Naturality Index equal to or lower than 2.5), such as plantations, reforestation and mountain forests, show the highest mean damage per year.In other forest types, I. typographus damage is instead mitigated by either the high diversity of the spruce mixed forests or the high elevation of the alpine forests.Damage caused by I. typographus was hence driven mainly by forest diversity, i.e., forest composition, and elevation, which directly affect tree resistance to insect colonization and indirectly affect insect population dynamics regulating the availability of susceptible host material.Results from this study clearly show that in forests growing at similar elevations, I. typographus infestations were largely influenced by forest composition, with mixed forests being less susceptible to bark beetle damage than pure spruce forests.This may be the result of several factors, which, however, often concern the physiological conditions and defenses of the potential host tree.Preformed or constitutional resin released upon attempted penetrations of the bark is, for instance, one of the main spruce defense mechanisms against I. typographus [37].In this respect, trees growing in mixed stands have been reported to have a higher resin flow than spruces from pure stands, increasing tree defense reactions against bark beetles and their associated fungi [38].Multi-tree species forests may also affect bark beetle chemical communication.Non-host volatiles, and, particularly, green leaf volatiles, released by broadleaves are known to reduce both I. typographus breeding performance [39] and its ability to find suitable host trees in mixed forests [40,41].Similar results were found by Hlá sny and Turčáni [3] in Central Europe.These authors suggest that the large infestations recorded in many European countries in the last few decades can be explained by the species composition of the monitored forests.Tree species composition is also an important predictor in bark beetle hazard rating models, such as Predisposition Assessment Systems [4].In this respect, an ecologically sustainable strategy to reduce the risk of I. typographus damage, especially in managed spruce forests, should be focused on forming stands with heterogeneous composition, reducing the proportion of spruce [11].A low amount of suitable breeding material, i.e., spruce, may prevent pests from building up their populations [42], whereas non-host trees may provide physical and chemical barriers to host location or colonization.Accordingly, our results show a mean number of infestation spots per year negatively correlated with the Naturality Index of the monitored forest types, with damage higher in pure (low index) than in mixed (high index) spruce forests.
Not just forest composition may affect the intensity of I. typographus damage.Within pure spruce forests, different forest types differently suffer I. typographus infestations, with spruce alpine forests growing at a high elevation (1300-1700 m a.s.l.) showing the lowest damage.A recent study by Chinellato et al. [43] explores how elevation, as a proxy for temperature, affects the abundance and diversity of bark and wood boring beetles associated with Norway spruce along the southern edge of the Alps.The paper shows that aggressive bark beetle species, such as I. typographus, respond positively to higher temperature, infesting spruce forests located at low elevations, i.e., warmer, more intensively than alpine forests [43].Growing in sites frequently exposed to unfavorable climatic conditions may reduce tree defenses against both climate disturbances and insect colonization [44].
Recent studies from northern [13] and central Europe [3] report that the availability of breeding substrates and the ability of I. typographus to colonize spruce trees growing in suboptimal conditions are the main abiotic factors triggering I. typographus outbreaks.A number of environmental factors may indirectly affect the population dynamics of I. typographus, acting on the susceptibility of the host trees through severe climate events, e.g., drought, heat waves, wind or snow storms, occurring in sites unsuitable for spruce growing [11,17,19].In this respect, the warm summer temperature recorded in the last few years in southern Europe [45] became one of the main factors affecting the general health conditions of spruce forests [46].Long, warm and dry summers may enhance tree susceptibility to forest pests [11,17,19], and I. typographus outbreaks were often associated with these events [11].In the same region where this study was carried out, Marini et al. [19] found an altitudinal distribution of I. typographus damage correlated with summer temperature and insect outbreak dynamics.Warm temperatures occurring at low elevations make spruce forests particularly vulnerable to I. typographus outbreaks [3,19].The same mechanism was observed worldwide also in other conifer bark beetles [8].Warm temperatures occurring at low elevations affect I. typographus also directly.In general, I. typographus populations living along the Southern Alps are bivoltine at a low altitude (<1300-1400 m a.s.l.) [15,18].Warm temperatures and long summers occurring at low altitudes allow high rates of development, with an increase in the number of generations per year, as suggested by several prediction models [47][48][49][50].Eruptions of bark beetle populations have often been associated with change in voltinism.Increased outbreaks of the spruce beetle, Dendroctonus rufipennis (Kirby), recorded in Alaska in the 1990s [51] have been, for instance, associated with a shift from a two-year to a one-year developmental cycle [52,53].Faccoli and Bernardinelli [18] recently investigated the breeding performance of the second generation in some bivoltine populations of I. typographus in the south-eastern Alps, suggesting that climate factors deeply influence the number of suitable hosts available to the second generation and, hence, the population growth.In conclusion, pure spruce forests growing at low elevations (<1300 m) suffer high bark beetle damage, because of warm temperatures stressing the trees, reducing host resistance to insect colonization, and allowing for a full second generation of the I. typographus populations.Finally, our results show that all considered forest types have infested trees, i.e., spruce diameter, and infestation spots with a similar mean size, suggesting that the total damage suffered by different spruce forest types is affected mainly by the number and frequency of the infestation spots rather than their size.In other words, once an I. typographus infestation begins, usually triggered by climatic events stressing spruce, its natural development proceeds independently from the forest composition and may terminate several years later following negative feedback mechanisms, such as intraspecific competition and natural enemies [11,21].In this respect, the probability of new infestations becomes the crucial point to assess the forest's risk of suffering I. typographus damage.A similar mechanism was reported for populations of the pine engraver beetle, Ips acuminatus (Gyllenhal), occurring in the south-eastern Alps [54], where, in years of high population density following drought periods, the number of infestation spots, but not their size, was significantly higher than in years of low density.

Conclusions
The recent intensification of bark beetle outbreaks recorded in European spruce forests [55] calls for a new management strategy, with a progressive reduction of pure spruce forests, especially at low elevations, and an increase in their ecological stability [3].Spruce should be mixed with tree species better adapted to the specific site characteristics and the new climatic conditions expected for the future [56].Increasing forest diversity by the introduction of broadleaf species will improve resistance to bark beetle infestations [57] and will assure the production of sustainable ecosystem goods and services [3].Forest protection measures should therefore be allocated according to the rating of the stand hazard based on site and stand characteristics, which vary among forest types.Within the perspective of a general improvement of tree health conditions, forest management should, moreover, shift from the stand to the regional scale, progressively reducing pure spruce forests or improving their general growing conditions.In conclusion, considering the potential increased susceptibility of spruce forests to I. typographus outbreaks due to climate change, there is increasing value in mitigating these effects through sustainable forest management, which includes the reduction of pure and intensive spruce cultivations and the improvement of their growing conditions [19].

Figure 2 .
Figure 2. Mean annual damage recorded in different forest types (m 3 every 1000 ha).Different letters indicate significant differences at p < 0.05.For the code legend, see Table1.

Figure 3 .
Figure 3. Mean annual number of infestation spots recorded in different forest types every 1000 ha of spruce forest.Different letters indicate significant differences at p < 0.05.For the code legend, see Table 1.

Figure 4 .
Figure 4. Mean size (m 3 ) of the infestation spots recorded in different spruce forest types.For the code legend, see Table1.

Figure 5 .
Figure 5. Correlation between the Naturality Index of different spruce forest types and the mean number of infestation spots suffered annually.