Since the 1950s, wind has represented one of the main disturbances to forest ecosystems in Europe, causing an increase in the frequency and amount of trees uprooted or broken by wind. Such phenomenon has also increased the incidence of compression wood in the xylem of remnant trees, thus leading to a general decrease in timber quality. In this study, we aimed to determine how wind regime affects the incidence of compression wood by reconstructing its chronology at both inter- and intra-annual scale using dendroclimatic methods. Six silver fir stands at different elevations were selected in central Romania, and compression wood time series were obtained from both increment cores from standing trees and radial discs from felled trees. Wood-structure variables were statistically analyzed as time series, while fluctuations of wind frequency and speed over the period 1921-present were reconstructed based on meteorological data. The structural response of sampled trees to wind was assessed based on the annual fluctuation in the frequency and intensity of compression wood. Results showed that the incidence of compression wood in the time series was synchronized with the intensity of the wind, rather than its duration. Wind regime in December of the preceding calendar year was significantly correlated with the frequency of compression wood, whereas its intensity was significantly correlated with wind load of previous September. The response of cambium to the wind stimulus increased with distance from the tree collar, peaking in the section at the base of the crown. At a decennial scale, only high-intensity wind stress triggered structural responses in the studied trees. Wind effects on xylogenesis in the analyzed stands increased over the last decades as a consequence of the local forest management. A better understanding of the structural response of forest trees to wind regime may explain how individual and groups of trees compete for stability and can help to improve forest management strategies in windy regions.
The frequency of intermittent or episodic strong winds has increased since the 1950s (
Plants exhibit phenotypic plasticity to wind, due to their almost continuous interaction with the moving air (
Dendrochronological analyses have demonstrated the utility of compression wood in reconstructing the wind-load history of trees (
Dendroclimatic studies on compression wood are quite rare, being mainly restricted to areas characterized by strong winds (
The study was carried out in the pre-mountainous region of the Postavaru Mountains, near Brasov (central Romania -
Silver fir (
The studied stands were uneven-aged or two-storey beech-silver fir mixed forests intensively managed by selective cutting, which was enforced after 1961. Harvest intensity varied in time and from one area to another.
We established six sampling plots (TR1, TR2, TR3, TR4, TR5, TR6) over a wide range of vegetation conditions (
We selected a total of 66 individual trees (
Two increment cores were extracted at the breast height from each sampled tree. One core was extracted in the direction of the prevailing wind, as inferred from the trunk geometry, while the second was extracted orthogonally to the first (
Sixty-three discs were collected in the same period from 25 trees felled in the harvested blocks near the TR2, TR3, and TR7 sample plots. Felled trees were crosscut at different distances from the collar (0, 4, 8, 12, 16, 20, 28 m -
The frequency of occurrence and the intensity of compression wood were examined in the sampled material. Two numerical-chronological series were used for each increment core and for each disc radius. The presence/absence of defects (compression wood) in a specific year along the series was recorded as 1/0, respectively. The intensity of the compression wood was assessed based on color and thickness of latewood, using the following numerical codes: (0) normal wood (the latewood was more yellowish than earlywood); (1) mild compression wood (the latewood ranged from light orange to red-yellowish); (2) moderate compression wood (latewood color was brown-yellowish-reddish); (3) severe compression wood (latewood color ranged from brown-reddish-yellowish to dark brown, spanning more than 50% of the width of growth rings -
To generate the axial growth-ring series (TROY), data from the series obtained for individual radii on each disc were assembled in mean series over each disc, which in turn were assembled in mean series over each height sector. For quantification purposes, the four measured radii formed a unique binary series, in which value 1 indicated the annual presence of a defect on at least one of the radii. The mean series of
Measurement of growth rings, the development of raw series of the growth rings, and their cross-dating were performed using the WinDENDRO® image-analysis system, the Density 2006c version (
Because a strong intraserial dependency was observed, the AutoRegressive Integrated Moving Average (ARIMA) model was used for the standardization of raw series (
Time series were prepared for standardization by smoothing. Negative-exponential or natural-logarithmic smoothing were used, and the series that were resistant to smoothing models were power-transformed or smoothed using the mobile moving-average method.
Meteorological records were obtained from the meteorological station in Brasov (45° 40′ N, 25° 37′ E, elevation 528 m a.s.l.), which is located in the vicinity of the sampled plots. Wind, temperature, and precipitation data from 1921 to present were used for the purposes of this study. Wind direction and intensity (speed) was recorded four times a day using an aerovane mounted 10 m above the ground. Wind indices were calculated from raw data and time series were constructed based on annual and monthly mean index values. Wind regime was defined using the following indices:
The values of these indices (hereafter, wind variables) were used to construct wind series. The difference in length of the series reflects the different number of observations in period considered and the improvement of recording instruments over time.
Structural indices (
The fluctuation in the values of structural indices accounted for by environmental factors was verified using dendrochronological statistics of the signal strength (
The relationship between wood-structure variables and climatic predictors was verified using the Pearson’s product-moment correlation coefficient (
All statistical analyses were carried out using the software package STATISTICA® v. 12.0 (StatSoft Inc, Tulsa, OK, USA).
Significant differences between consecutive growth rings belonging to average non-detrended series (TROI and TROY) were found using the
A preliminary examination of the internal variability of structural indices was performed using the dendrochronological statistics MS, ESR, and PACF (
The time series of wind-related variables (
Cluster analysis of wood structural indices for the trunk base sector revealed a discrepancy between the
The ability of the wood-structure time series to reflect fluctuations in wind-related variables is shown in
The strongest correlations with the wind explanatory variables (
Axial series showed stronger individual correlations with wind variables, that can be summarized as follows: (i) the basal trunk sector had the highest partial correlation between
The axial series (TROY) were found to better reflect monthly fluctuations of wind variables compared to core series (TROI -
The use of the TROY-7 series increased the partial correlation coefficient between
Regarding the different trunk sections, we noticed that the mild compression wood fractions in the extreme sectors of the bole (1 and 7) were related to the relatively mild winds of June and February. As for the intermediate height sectors (especially the second sector: 4-8 m), the intensification of wind in March and May was related to the incidence in compression wood (
The analysis of the chosen time series at a decennial scales revealed a stronger relationship with annual fluctuations of wind regimes as compared with the full time series (
The first major reduction in stand density due to human interventions occurred in 1969 in sample plot TR6, when trees were 50-150 years old. This density reduction resulted in a larger correlation coefficient between
It has been reported that peaks in compression wood follow with a certain lag time the peaks in solid precipitation and windstorm events (
The frequency of windthrow events noticeably increased after 1970 throughout Europe (
The exploration of historic wind regime at the annual resolution showed that wind perturbations are influenced or accompanied by a long-term intensification of air-mass movement. Our results proved that the compression wood is synchronized with the intensity of the wind, rather than with its duration. This study demonstrated that the analysis of the radial growth is an accurate tool in the study of interactions of silver fir populations with moving air masses (
Given that tree stability is conditioned by its underground architecture (
The major wind events (based on fluctuations in
The variation of the incidence in compression wood along the trunk (
The
In this study, 61% of the explored series failed to explain the effect of climatic variables on the incidence of compression wood, suggesting the existence of different sources of variation. A continuous improvement over time of the relationship between the incidence of compression wood and wind-related variables at a decennial scale was observed in the TROI-E and TROI-F time series (
Stand thinning increased the incidence of
The instability of the correlation between
Despite the decline in the frequency of strong windstorms in the last 150 years (
This study demonstrates that wind affects radial growth of silver fir trees, even in habitats where wind is not a limiting factor for growth. Peaks of incidence in compression wood were synchronous with increases in air turbulence over the period 1921 to present.
The non-detrended time series of compression wood showed a high degree of internal dependence, which emphasized the persistence of climatic excitation in the compression wood records and led to the analysis of the previous calendar year contributions. The analysis of standardized series showed the influence of wind conditions in the preceding year on the incidence of the compression wood, while its intensity is better correlated with the wind regime in the current year.
The response of trees to fluctuations in wind conditions varied in space, even at relatively low distances. Trees growing at higher elevation showed a stronger wind effect in the time series of compression wood. At the lowest altitude, wood compression may be explained by the weakening of tree anchoring in hydromorphic soils.
Variation of the incidence of compression wood along the trunk suggests that cambium sensitivity to wind-induced mechanical stress varies or the transduction of wind load in woody tissue changes in different part of the stem. Based on our results, the trunk section at the base of the crown (16-20 m) better reflects annual wind movements, while the incidence of compression wood at height of 4-8 m reflects monthly wind circulation.
On a monthly scale, wind conditions in previous December, current March and current May stimulate the occurrence of compression wood, whereas those in previous September, current February and current May affect its intensity.
The analysis of residual chronologies of compression wood at the decennial scale showed that fluctuations in cambium capacity reflect long-term wind conditions, and revealed that wind strongly contributed to the formation of compression wood up to a decade after human interventions.
The following abbreviations have been used throughout the text:
This work was supported by the Romanian Executive Agency for Higher Education, Research, Development and Innovation Funding (Project number TD-79: “Structural macroscopic features of raw wood as a means of estimating its quality”). We thank Eng. Laurentiu Puscasu for assisting with the field work. We also thank the two anonymous referees for providing constructive remarks.
Geographic location of the study site (upper panel) and distribution of the sample plots over the study area (lower panel).
Prevalent wind directions in the study area (multiannual values over the 1964-2004 period). Data source: Brasov meteorological station, Romania (45° 40′ N, 25° 37′ E, elevation 528 m a.s.l.)
Dendrogram of mean chronologies of structural indices obtained from the base portion of the trunk.
Pearson’s correlation coefficients between wind-related variables and structural indices in the detrended time series analyzed. (a): yearly calm frequency (
Partial correlations between compression wood residual chronologies (after removing the influences of temperature and precipitation) and wind variables in (a) core series (TROI) and (b) disc series (TROY). (
Effect of the wind seasonal regime on the annual frequency of compression wood (
Wind signal in compression wood fluctuations at a yearly (a, c) and decennial (b, d) scales.
Main environmental characteristics of the sample plots. (*): Soil classification according to
Plot | Long | Lat | Elev. (m) | Aspect | Slope (°) | Soil type* | APL |
---|---|---|---|---|---|---|---|
TR1 | 25°37′ E | 45°37′ N | 650.7 | - | 2 | Dystric Gleysol | Low |
TR2 | 25°37′ E | 45°36′ N | 697.3 | SE | 27 | Stagni-albic Luvisol | Increased |
TR3 | 25°36′ E | 45°37′ N | 765.5 | NV | 16 | Dystric Cambosol | Moderate |
TR4 | 25°36′ E | 45°36′ N | 819.3 | V | 26 | Dystric Cambosol | Increased |
TR5 | 25°36′ E | 45°36′ N | 948.4 | NV | 36 | Lepti-dystric Cambisol | Moderate |
TR6 | 25°36′ E | 45°37′ N | 798.3 | S | 25 | Dystric Cambosol | High |
Tree sample size and labels of the time series analyzed in this study. (TROI): increment core series; (TROY): axial growth-ring series.
Sampledtrees | Source plot /Height sector | Growth-ring series label | |||
---|---|---|---|---|---|
Standing trees | TR1 | TROI-A | 13 | 25 | 883 |
TR2 | TROI-B | 7 | 14 | 1361 | |
TR3 | TROI-C | 9 | 18 | 1831 | |
TR4 | TROI-D | 12 | 15 | 1128 | |
TR5 | TROI-E | 12 | 17 | 1774 | |
TR6 | TROI-F | 13 | 22 | 1962 | |
- | TROI (full core’s dataset) | 66 | 111 | 8939 | |
Felled trees | 0-3.9 m | TROY-1 | 6 | 24 | 3425 |
4-7.9 m | TROY-2 | 7 | 26 | 3623 | |
8-11.9 m | TROY-3 | 12 | 47 | 5286 | |
12-15.9 m | TROY-4 | 7 | 27 | 2354 | |
16-19.9 m | TROY-5 | 11 | 43 | 2982 | |
20-27.9 m | TROY-6 | 10 | 40 | 2703 | |
≥ 28 m | TROY-7 | 10 | 40 | 1479 | |
- | TROY (full disc’s dataset) | 63 | 247 | 21852 |
Descriptive statistics of the non-detrended growth-ring average series. (Avg): average; (CV): coefficient of variation.
Structuralindex | Statistics | Series | |
---|---|---|---|
TROI | TROY | ||
Commoninterval | 1825-2007 | 1866-2006 | |
Avg | 1.790 | 1.977 | |
Max | 3.139 (1977) | 3.541 (1914) | |
Min | 0.703 (1849) | 0.977 (2003) | |
CV | 31.92 | 24.03 | |
Avg | 0.683 | 0.617 | |
Max | 1.134 (1998) | 1.092 (1881) | |
Min | 0.232 (1849) | 0.286 (2003) | |
CV | 31.93 | 26.35 | |
Avg | 35.98 | 32.93 | |
Max | 44.30 (1899) | 43.08 (1880) | |
Min | 24.28 (1853) | 23.55 (1928) | |
CV | 8.98 | 9.83 | |
Avg | 22.58 | 11.76 | |
Max | 69.50 (1907) | 34.43 (1880) | |
Min | 4.17 (2006) | 0 (1860, 1861, 1863-1865) | |
CV | 52.68 | 52.43 | |
|
Avg | 1.85 | 1.72 |
Max | 3.00 (1902) | 2.33 (1953) | |
Min | 1.05 (1985) | 1.17 (1986) | |
CV | 22.14 | 14.67 |
Signal properties of the examined time series. (SI): Structural index; (SS): Status of standardization; (DS): Dendrochronological statistics; (*) first-order autocorrelation.
SI | SS | DS | Core growth-ring series (TROI) | Axial growth-ring series (TROY) | |||||||||||
---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
TROI-A | TROI-B | TROI-C | TROI-D | TROI-E | TROI-F | TROY-1 | TROY-2 | TROY-3 | TROY-4 | TROY-5 | TROY-6 | TROY-7 | |||
|
Beforedetrending | MS | 0.145 | 0.160 | 0.130 | 0.137 | 0.129 | 0.141 | 0.178 | 0.156 | 0.137 | 0.165 | 0.153 | 0.153 | 0.185 |
ESR | 0.597 | 0.759 | 0.747 | 0.770 | 0.806 | 0.693 | 0.813 | 0.776 | 0.682 | 0.760 | 0.660 | 0.664 | 0.740 | ||
ACF_1* | 0.815 | 0.921 | 0.867 | 0.838 | 0.932 | 0.767 | 0.827 | 0.823 | 0.869 | 0.877 | 0.719 | 0.774 | 0.329 | ||
ARIMAchronologies | MS | 0.245 | 0.269 | 0.245 | 0.307 | 0.234 | 0.236 | 0.266 | 0.241 | 0.220 | 0.213 | 0.268 | 0.221 | 0.264 | |
REFF | 0.560 | 0.331 | 0.038 | 0.158 | 0.085 | 0.237 | 0.446 | 0.371 | 0.286 | 0.181 | 0.200 | 0.327 | 0.377 | ||
EPS | 0.943 | 0.776 | 0.299 | 0.674 | 0.527 | 0.802 | 0.829 | 0.805 | 0.828 | 0.607 | 0.733 | 0.829 | 0.858 | ||
SE | 0.184 | 0.309 | 0.327 | 0.277 | 0.276 | 0.242 | 0.304 | 0.300 | 0.244 | 0.342 | 0.270 | 0.259 | 0.250 | ||
|
Beforedetrending | MS | 0.189 | 0.230 | 0.175 | 0.192 | 0.199 | 0.187 | 0.265 | 0.225 | 0.182 | 0.247 | 0.231 | 0.266 | 0.257 |
ESR | 0.554 | 0.785 | 0.618 | 0.660 | 0.693 | 0.581 | 0.810 | 0.740 | 0.603 | 0.729 | 0.635 | 0.684 | 0.702 | ||
ACF_1 | 0.779 | 0.814 | 0.733 | 0.756 | 0.865 | 0.605 | 0.663 | 0.641 | 0.672 | 0.671 | 0.328 | 0.611 | 0.248 | ||
ARIMA chronologies | MS | 0.354 | 0.419 | 0.303 | 0.331 | 0.042 | 0.352 | 0.082 | 0.064 | 0.081 | 0.078 | 0.076 | 0.109 | 0.068 | |
REFF | 0.423 | 0.263 | 0.067 | 0.121 | 0.151 | 0.139 | 0.236 | 0.275 | 0.246 | 0.201 | 0.190 | 0.295 | 0.431 | ||
EPS | 0.905 | 0.714 | 0.393 | 0.602 | 0.681 | 0.677 | 0.650 | 0.659 | 0.797 | 0.638 | 0.721 | 0.807 | 0.883 | ||
SE | 0.211 | 0.324 | 0.322 | 0.283 | 0.266 | 0.257 | 0.357 | 0.348 | 0.251 | 0.338 | 0.271 | 0.266 | 0.239 | ||
|
Beforedetrending | MS | 0.514 | 0.704 | 0.220 | 0.625 | 0.423 | 0.401 | 0.673 | 0.558 | 0.213 | 0.543 | 0.537 | 0.373 | 0.465 |
ACF_1 | 0.671 | 0.718 | 0.765 | 0.728 | 0.805 | 0.692 | 0.643 | 0.718 | 0.610 | 0.628 | 0.507 | 0.752 | 0.497 | ||
|
Beforedetrending | MS | 0.174 | 0.156 | 0.164 | 0.116 | 0.142 | 0.127 | 0.118 | 0.152 | 0.182 | 0.158 | 0.176 | 0.143 | 0.261 |
ACF_1 | 0.667 | 0.586 | 0.649 | 0.777 | 0.719 | 0.735 | 0.748 | 0.643 | 0.428 | 0.698 | 0.536 | 0.624 | 0.075 |
Descriptive statistics of the non-detrended time series of wind-related variables. (Avg): average; (ME): median; (MAvg): multiannual average; (CV): Coefficient of inter-annual/ inter-monthly variation.
Windvariable | Commoninterval | Statistics | ||||
---|---|---|---|---|---|---|
Avg | ME | Range | CV(%) | |||
Min (year/month) | Max (year/month) | |||||
|
87(1921-2007) | 35.9 | 38.9 | 2.0 (1954, 1955) | 65.6(1951) | 40.9 |
|
44(1964-2007) | 43.7 | 43.8 | 34.8(April MAvg) | 53.7(January MAvg) | 28.9 |
9.2(July 2006) | 94.9(December 1994) | |||||
|
67(1941-2007) | 3.5 | 3.4 | 1.6(1949) | 5.4(1998) | 36.6 |
|
44(1964-2007) | 4.3 | 4.3 | 3.5(February MAvg) | 4.9(March and November MAvg) | 30.5 |
0.3(August 1984) | 9.2(January 1976) | |||||
|
87(1921-2007) | 15.7 | 16.0 | 0(1944) | 63(1974) | 69.7 |
|
35(1973-2007) | 2.8 | 2.0 | 0(1987, 1989, 1996, 1997, 1999, 2004-2007) | 11(1981) | 109.6 |