Tree-ring widths have globally been used for investi-gating the impact of environmental factors on tree-growth dynamics such as the rising atmospheric CO2 concentration (Gedalof and Berg, 2010) and rapid changes in SO2 and NOx air pollution (Bošeľa
The reliability of tree-ring widths (TRW) relies on the exact dating of the year of their formation and therefore the correct dating of tree rings is a crucial methodological step in all dendroecological and dendroclimatic studies (Fritts and Swetnam, 1989; Maxwell
The issue is particularly important for diffuse-porous and semi-ring-porous tree species such as European beech wherein the vessel-size distribution is uniform throughout the year creating tree rings that are difficult to read (Schweingruber, 1988; DeRose and Gardner, 2010). European beech was found to have high a potential for dendrochronological studies (Dittmar
Due to a high degree of uncertainty with the cross-dating process, checking the dating accuracy against independent reference chronologies is highly desirable. With beech, the difficulty often emerges from the fact that as a widespread shade-tolerant tree species it frequently grows in unfavourable environments and/or be-low the canopy layer of dominant coexisting tree species (e.g. oak, spruce). At the upper distributional limit of beech, cool years with short growing seasons or extreme frost events could lead to the formation of very narrow or even missing rings (Hantemirov
In an effort to allow the full utilization of the dendro-chronological potential of beech and to contribute to improving the cross-dating process in general, we present a technique where visual dating before measurement is validated by instrumental climate data dated with absolute certainty. The properties and practical applicability of this method are demonstrated on datasets cross-dated without an independent regional chronology.
The study area is situated in the upper altitudinal vegetation belt (1300–1450 m a.s.l.) of the Poľana Mountain (48°38'N, 19°29'E), which is part of the Slovak Ore Mts. and belongs to the Western Carpathians. The area is char-acterized by cold mountain climate with a mean annual temperature of 3.5–4.0°C and an annual precipitation sum of 900–1100 mm. Here, European beech reaches its upper limit and grows mostly under the canopy of Norway spruce
The study material was collected in the years 2012 and 2013 at two sites near the top of the Poľana Mt. (Fig. 1). The site Predná Poľana (48°37'46''N, 19°27'44''E - PP) is characterized by less dense spruce-dominated stands overgrowing an abandoned pasture at 1300 m a.s.l. The prevailing south and south-east exposition enables beech ascending to high altitudes in sparse spruce forests.
The site Zadná Poľana (48°28'23''N, 19°29'05''E - ZP) is characterized by a very old closed-canopy primeval spruce forest. The prevailing expositions are east and south-east. Beech occurs sporadically between 1100 and 1200 m a.s.l., i.e. 100–200 metres lower than at the PP site, mostly in patches of subdominant and dominant trees.
Monthly observations of CRU TS3.21 Mean Temperature (0.5° × 0.5° grid interpolated points) covering the time period since 1901 available at KNMI Climate Explorer (Harris
Eleven trees (two cores per tree) were sampled at Predná Poľana (dataset PP) and 10 trees (one core per tree) were sampled at Zadná Poľana (dataset ZP). The trees at PP were younger (pith age 60–80 years) than the trees at ZP (pith age 130–170 years). All cores were mounted into 5 mm wooden bars, air-dried and sanded by grain paper (400–600 grain). The TRW were measured to the nearest 0.01 mm with the digital positiometer DAS (Digital Analysis System, Janíček, 1994) and processed by standard methods (Cook and Kariukstis, 1990) using the TSAP-Win software (Rinn, 2003). In addition, at PP, the changes in stem circumference were measured using automatic band dendrometers — DRL 26 (EMS Brno, Cz), installed on three beech trees at 2.5 m above the ground during seven vegetation periods from 2007–2013.
Three cross-dating methods were applied: (1) Visual cross-dating by graphical comparison and alignment of individual TRW series (Schweingruber, 1983); (2) dating by skeleton plotting in combination with the memorization method when the visual quality of the study material did not permit the use of simple graphics (Speer, 2010); (3) a new procedure for cross-dating of cores from living trees (referred to as combined dating method) (see Appendix A).
The practical applicability of the combined dating method was tested and validated in separate ways: (i) comparing the dating results by two independent re-searchers; (ii) comparison with an independent reference chronology for beech growing at its upper limit in the nearby mountain range Vtáčnik (48°38´N, 18°39´E; 1240 m a.s.l.) (Dittmar
Cross-dating of a problematic core using the combined dating approach was verified by micro-sectioning and examination under a light microscope. Thin sections were prepared following Gärtner and Schweingruber (2013). The core was split length wise into 2 parts. The part used for detailed examination was saturated in water and complete cross-section (approx. 20 µm thick) was cut using a GSL1- microtome (Gärtner
Tree-ring series in a pure beech stand (site Veľké Pole-dataset VP; 48°33.29'N, 18°31.10'E; 500–530 m a.s.l.) in the Štiavnické Mts. were collected from 13 trees in the canopy layer in 2014 (one core per tree either from the north or east exposed side). This 50-year old beech stand down to the east and with an slope of 30° is located at the boundary between the optimum and the lower distributional limit of beech. Monthly observations of CRU TS3.21 Precipitation (0.5° × 0.5° grid interpolated points) since 1901, available at KNMI Climate Explorer (Harris
Dendrochronological characteristics such as mean TRW, standard deviation (SD), mean sensitivity (MS), first-order autocorrelation (AC1), series inter-correlation (Rbar), coefficient of coincidence (Gleichläufigkeit GLK%) and expressed population signal (EPS) were calculated for the raw site chronologies using the average of the TRW series at each site after cross-dating (Table 1). Both higher MS and SD values indicate that tree-ring series are very sensitive, i.e. they exhibit a large inter-annual variability of TRW, a higher frequency of irregu-lar growth patterns, completely missing rings or extreme-ly narrow rings (micro-rings) that may hamper the cross-dating. Low AC1 values indicate that the series are not highly auto-correlated which could help facilitate the cross-dating process. Despite a higher sensitivity of individual series and their low number, the rather high values of GLK%, EPS and Rbar indicate successful cross-dating. The resulting chronologies are of good statistical quality and are representative for a common climatic signal rec-orded for a site (tree species population) in the mean chronology (sensu Wigley
Basic descriptive statistics of the raw European beech tree-ring site chronologies.
Dataset PP — site Predná Poľana, ZP — site Zadná Poľana, VP — site Veľké Pole | Period Period and length of chronology with sample depth over 4 series; TRW — tree-ring width (1/100 mm); N — number of series, in parenthesis mean number of series in chronology; SD — standard deviation; AC(1) — 1st order autocorrelation; GLK% — coefficient of coincidence Gleichläufigkeit; MSc% — mean sensitivity of the chronology; MSs% — mean sensitivity of individual series in %; R-bar — series inter-correlation; EPS — expressed population signal. | TRW Mean Min-Max | N | SD | AC(1) | GLK% | MSs% Mean Min-Max | MSc% | R-bar | EPS |
---|---|---|---|---|---|---|---|---|---|---|
PP | 1940–2012 | 247 | 11 | 71.5 | 0.35 | 79 | 34 | 27 | 0.48 | 0.90 |
73 | 91–383 | (10) | 26–53 | |||||||
ZP | 1857–2012 | 148 | 10 | 50.4 | 0.56 | 77 | 38 | 27 | 0.56 | 0.92 |
156 | 45–302 | (9) | 34–50 | |||||||
VP | 1957–2014 | 210 | 13 | 52.4 | 0.21 | 82 | 31 | 26 | 0.48 | 0.92 |
58 | 78–328 | (13) | 26–37 |
Increment cores from the PP dataset were cross-dated using the graphical comparison method (Schweingruber, 1983). The tentative reference curve was constructed as the arithmetic mean of the individual TRWs for each calendar year from a subset of seven series with the highest interseries correlation (r > 0.6) and the highest coefficient of coincidence (GLK > 80%). The rest of the material was graphically cross-dated with the tentative reference curve. Skeleton plotting, in combination with the memorization method (Speer, 2010), was used for the dataset ZP, in which most cores showed difficult to recognize ring structure and rather poor quality. The tentative reference curve was also constructed from a subset of the seven best measurable series. The comparison of the two reference curves (constructed separately by two in-dependent operators in 2012 and 2013) from two nearby localities showed several apparent inconsistencies in dating of the same marker rings and characteristic ring patterns (visually clearly detectable in both datasets). When the cross-dating of the individual series in both ZP and PP data sets were checked with COFECHA (Holmes, 1983; Grissino-Mayer, 2001), no crucial errors or misdated series were identified.
As no local tree-ring chronologies were available as independent references, instrumentally measured climate records were taken to check the calendar years of tree-ring formation. We assumed that temperature is the most limiting factor for the growth of beech at its upper distribution limit. According to Ježík
Comparison of the mean curve constructed by the combined dating method and the mean curve constructed by the graphical method for the PP site (Fig. 4a) shows that both methods provide consistent results for the last 40–50 calendar years. Correlations between the tempera-ture records and common graphical method suddenly drop and are consistently lower in comparison with the moving correlations of the combined dating method. The combined dating method also improves the empirical signal strength statistics of the mean chronology (Rbar increases by 0.07 and EPS by 0.02).
To validate the combined dating method, another dendrochronologist independently applied the procedure to all 11 tree-ring series from the PP dataset (Fig. 4b). By visual examination of the wood, assigning and recording marker rings and prominent growth patterns the dendro-chronologist was able to construct a reference curve that coincides with our reference curve (no difference in Rbar and EPS statistics). Additionally, the reference chronology for the Poľana Mt. was checked for dating against the independent reference curve from the Vtáčnik Mts. (Dittmar
In addition, the evidence of the dating ability of the combined dating method was validated using light microscopy.
After inspection of the micro-section the number of rings were counted and measured. The comparison of an established chronology and the single tree-ring series revealed the problematic segment (Fig. 7a). The combined dating method indicated 8 missing rings confirmed by micro-sectioning (Fig. 7b). The possible explanation of this issue is that suppressed sensitive trees exhibited repeatedly reductions in growth lasting 5–8 years.
The core idea of the proposed cross-dating approach is the validation of uncertain subjective decisions in da-ting of the tree ring series. The method especially applies in cases of unclear TRW boundaries, frost rings, intra-annual growth pulses, ring wedging, missing or micro-rings,
The proposed cross-dating approach is to simplify and verify the subjective decision of the researcher regarding the visual assessment of tree-ring boundaries. In general, for each tree ring there is five possibilities affecting the dating (insert, join, delete, divide and leave without change). For example, in the 30-year segment with only, let’s say, 4 problematic boundaries there exist in total 54 = 625 combinations of possible decisions. A large sampling design is thus required to explicitly capture complicated spatial patterns of climate-growth relation-ships even for relatively small geographic areas (Büntgen
Extensive research is needed especially for the shade-tolerant beech as one of the main tree species in Europe, which often grows in understorey of the mixed stands on boundaries of its natural range (i.e. climate sensitive sites). Although some reference chronologies were developed for conifers in the region, their climate-growth responses often differ from that of broadleaved trees, even in the same locality (Castagneri
Under such circumstances it is very convenient to use freely or easily available instrumental records as an alter-native reference chronology. One option can be using the climate records from the nearest meteorological stations. However, the proposed method does not require very precise climate data for the each sampling point. As was shown in our study, gridded CRU TS, a freely available climatic series, was sufficient for cross-dating and has already been used in many dendroclimatological studies in other European regions (e.g. Babst
The instrumental time series is used to control the subjective decisions of the researcher. The climate records partly replace the role of working reference chronol-ogy (which cannot be reliably constructed using a very small sample size) and partly serve as a validation tool similar to an independent reliable dated regional chronol-ogy (which might not be available for the site). In gen-eral, the climate records mainly serve for the identifica-tion of suspicious segments with reduced climate-growth correlation, suggesting a misdating, which should be re-assed and re-inspected on the wood.
However, a concern of a “cycling error” (using climate records for validation of tree-ring sequences and then using these sequences for climate reconstructions) could arise. But indeed, this possibility is almost excluded if two simple conditions are met: (i) any purposive adjustment of the measured tree-ring widths in order to artificially increase the correlations with climate without visual justification on the wood are not allowed and per se (ii) manipulation with available climate records is fully excluded. When trial shifts of dating position of the segment does not result in a correlation increase, the low correlation with climate is accepted as a fact and any deliberate interventions to TRW measurements are not justified and thus not allowed. On the contrary, employing climatic records for cross-dating can be considered as the first phase in constructing a dendroclimatic model and can improve the extraction of the climate signal from tree-ring series.
Anyhow, the method is based on the assumption that the relationship between climate and tree growth has a non-zero magnitude and is time-stable (Fritts, 1976). Although significant and temporally stable correlations when testing temporal coherence between TRW, maximum latewood density (MXD) and observed JJA summer temperatures in chronologies from cold environments have been demonstrated (e.g. Esper
Adverse (cool, rainy and short) growing seasons in temperature limited environments may lead to the formation of very narrow rings (micro-ring, Hantemirov
Many of the problematic trees are suppressed trees for which the exclusion is acceptable when considering that unreliable cross-dating would introduce a lot of noise to the chronology (Lorimer
If core exclusion is unavoidable, the additional coring of more trees from particular or nearby localities should be employed. However, both approaches – microsectioning and/or additional coring – are rather costly and time consuming. From this point of view, the combined cross-dating method has a good potential for better exploitation of empirical material.
The proposed combined method resulted in the successful construction of a reference chronology despite the small sample size and the absence of the regional chronology. The proposed dating technique demonstrates its ability to markedly facilitate the process of cross-dating, particularly of trees from limiting environments or sup-pressed trees. The method allows the detection of rings that are completely missing or hardly identifiable.
The method was applied to ring series of living beech trees from sites where temperature or precipitation was identified as the most prominent growth limiting factor. The applicability of our approach thus needs further testing for different environments and tree species where other climate factors drive the tree growth. However, since the method has been successful for small datasets of under-canopy trees and problematic species (beech), the reasonable assumption exists that it will likely work for larger datasets that include species with better recognizable annual rings.
Up-to-date there is no published or generally accepted dating method that is fully independent of human bias. According to Cook and Kairiukstis (1990) “