Cell wall dimensions reign supreme: cell wall composition is irrelevant for the temperature signal of latewood density/blue intensity in Scots pine

(1)

Dendrochronologia 65 (2021) 125785

Available online 9 November 2020

(http://creativecommons.org/licenses/by-nc-nd/4.0/).

ORIGINAL ARTICLE

Cell wall dimensions reign supreme: cell wall composition is irrelevant for the temperature signal of latewood density/blue intensity in Scots pine

Jesper Bj orklund ¨

^a^,

*, Marina V. Fonti

^a^,^b

, Patrick Fonti

^a

, Jan Van den Bulcke

^c^,^d

, Georg von Arx

^a

aSwiss Federal Institute for Forest, Snow and Landscape Research WSL, Zürcherstrasse 111 CH-8903 Birmensdorf, Switzerland

bInstitute of Ecology and Geography, Siberian Federal University, Svobodny pr. 79, 660041 Krasnoyarsk, Russian Federation

cUGent-Woodlab, Laboratory of Wood Technology, Department of Environment, Faculty of Bioscience Engineering, Ghent University, Ghent, Belgium

dGhent University Centre for X-ray Tomography (UGCT), Ghent, Belgium

A R T I C L E I N F O Keywords:

X-ray density Dendroanatomy Blue reflectance Lignin content Palaeoclimatology

A B S T R A C T

Many microdensitometric techniques are available for deriving maximum latewood density (MXD), which is the state-of-the-art proxy parameter for local to hemispheric-scale temperature reconstructions of the last millen- nium. Techniques based on X-ray radiation and visible light reflection, such as “blue intensity” (BI), integrate both the density/composition and the dimensions of the cell walls to derive microdensitometric data. In contrast, the dendroanatomical technique relies only on the dimensions of the cell walls. It is therefore possible to isolate cell wall variables by subtracting data derived using the dendroanatomical technique from data derived using X- ray and BI-based techniques.

In this study, we explore differences in well-replicated data from parallel X-ray, BI, and dendroanatomical measurements of temperature-sensitive Pinus sylvestris trees from northern Finland. We aim to determine whether cell wall density is critical to the success of X-ray-based MXD, and whether the BI-based parameter counterpart, here termed MXBI, contains useful information about the composition of the cell wall (specifically the lignin).

Our results indicate that cell wall density and cell wall BI have no relevant influence on MXD and MXBI measurements. Even in years with severely reduced lignification, identified as so-called “blue rings”, dendroanatomical MXD (aMXD) measurements do not deviate significantly from their MXD or MXBI counterparts.

Moreover, derived chronologies of cell wall density and cell wall BI contain no significant climate signals when correlated with local climate. Maximum latewood density of conifers can thus be obtained without bias using the dendroanatomical technique. Because lignin content appears to play a negligible role for cell wall BI, the cell wall BI likely presents the biggest challenge when producing unbiased MXBI data. This is because BI data is notorious for cell wall color distortion across the heartwood and sapwood, and between living wood and dead wood, and may therefore distort the otherwise strong link with wood density on multidecadal scales.

1. Introduction

Maximum latewood density in tree rings (MXD) has obtained a unique position in Northern Hemisphere dendroclimatic studies due to the superior temperature information that can be gained from it as compared to that from tree-ring widths (e.g., Ljungqvist et al., 2020;

Wilson et al., 2017). However, its standing is somewhat clouded by the multitude of microdensitometric techniques that are available to derive various versions of MXD or MXD-related information (e.g., Bjorklund ¨ et al., 2019; Jacquin et al., 2017). It is therefore important to examine the basic measurement principles and output data of each of the various techniques to understand where differences lie and how to optimize

their applications. Potential “inter-technique” differences may even provide new and important perspectives regarding the utility of each technique.

Wood density (xylem density) is theoretically a product of the proportion of tracheid cell wall in an analyzed volume or surface, i.e., the dimensions of the cell walls divided by the dimensions of the wood sample (e.g., Vaganov et al., 2006; Elliott and Brook, 1967; Jagels and Telewski, 1990; Polge, 1978; Tsoumis, 1964). However, xylem density is modulated by the chemical composition of the solid cell walls, which in turn controls the density of the cell walls (e.g., Vaganov et al., 2006).

Xylem density is thus an aggregate of both the cell wall dimensions and cell wall composition. Cell wall dimensions are clearly the most

* Corresponding author.

E-mail address: jesper.bjoerklund@wsl.ch (J. Bj¨orklund).

Contents lists available at ScienceDirect

Dendrochronologia

journal homepage: www.elsevier.com/locate/dendro

https://doi.org/10.1016/j.dendro.2020.125785

Received 24 March 2020; Received in revised form 24 September 2020; Accepted 29 October 2020

(2)

important driver of MXD (e.g., Bj¨orklund et al., 2020), but it is not yet understood to what extent cell wall composition influences microdensitometric measurements intended for dendroclimatic purposes.

The tracheid cell wall consists of several distinct, intricately composed sub-layers (Butterfield, 2003) that are mainly comprised of cellulose, hemi-cellulose, and lignin macromolecules (Pereira et al., 2003). These macromolecules have slightly different densities (lignin ~ 1.35 g/cm³; hemicelluloses = 1.5–1.8 g/cm³; cellulose =1.56 g/cm³ (Saranpäa, 2003) and colors. Pure cellulose is white, whereas lignin ¨ turns yellow in air due to photo-oxidation (e.g., Brunow and Sivonen, 1975; Leary, 1968). Generally, cellulose is most abundant in the conifer cell wall (40–50 %), followed by hemi-celluloses with 20–35 % and lignin (15–35 %) (Sarapää, 2003). However, despite substantial variation in cell wall composition across various tree species (Saranpää, 2003), inter-specific cell wall density is most often found to be remarkably invariable at about 1.5 g/cm³(e.g., Hill and Papadopoulos, 2001). Although there is controversy regarding whether cell wall density increases, decreases, or stays the same intra-annually (from earlywood to latewood) (cf. Stamm and Sanders, 1966; Decoux et al., 2004;

Wilfong, 1966; Yiannos, 1964; Silkin and Ekimova, 2012), these studies largely agree that cell wall density is irrelevant to the change in intra-- annual xylem density. This suggests that it is unlikely that systematic inter-annual variability in cell wall density or composition would affect inter-annual xylem density in a meaningful way. However, if the packing density and/or relative abundance of each molecule class (i.e., cell wall composition) were influenced by year-to-year variability in envi- ronmental conditions, this would theoretically be reflected by variability in the density and light reflection of the cell walls.

Ascertaining the relevance of cell wall composition to inter-annual variability of xylem density would help resolve the question of whether important information is overlooked when novel dendroanatomical techniques are used to produce wood density data.

Dendroanatomical density data rely on cell wall dimensions and ignore the cell wall composition completely. If an important part of the inter- annual xylem density variability were driven by variations in cell wall composition, it would be important to try to isolate or separate this from the cell wall dimensions. Variables of cell wall composition may even provide new tree-ring proxy parameters for further exploration (sensu Silkin and Ekimova, 2012). In fact, the differences between most microdensitometric techniques and the dendroanatomical technique should represent technique-specific quantifications of cell wall density, light reflection, or composition. For example, the difference between X-ray density and dendroanatomical density should represent cell wall density (sensu Decoux et al., 2004; Silkin and Ekimova, 2012; Silkin and Kirdyanov, 2003). Similarly, the difference between “blue reflectance”, a.k.a. blue intensity (BI; McCarroll et al., 2002), and dendroanatomical density should represent cell wall BI. McCarroll et al. (2002), hypothesized that cell wall BI is closely related to lignin content. Similarly, Gindl et al. (2000) found that the inter-annual variation in latewood lignin content is correlated with late summer/fall temperatures. However, Gindl et al. (2000) also found that latewood lignin content is uncorre- lated with MXD. If the hypothesis of McCarroll et al. (2002) and the results of Gindl et al. (2000) are correct, derived cell wall BI should contain useful climate proxy information that is distinct from what can already be found in MXD. Moreover, Piermattei et al. (2015) found that certain unusual rings identified in double-stained micro-sections, called

“blue rings”, are blue in color as a result of markedly reduced lignification of the cell walls. A normally lignified cell wall should be colored red from the safranin. If the lignin content of the cell wall is an important component of BI data, it should be possible to detect substantial differences between dendroanatomical density and BI, at least when comparing “blue rings” with normally lignified rings. This would be an additional test of the hypothesis of McCarroll et al. (2002).

In this study, we explore whether variations in cell wall density and cell wall BI have important influences on MXD and the corresponding MXBI parameter from the BI technique. We use parallel X-ray, BI, and

dendroanatomical measurements of temperature-sensitive Pinus syl- vestris trees from northern Finland for the analysis. We further investi- gate whether variations in cell wall density and cell wall BI contain any useful climate information; if so, it may be possible to exploit these variations as novel proxy parameters. Finally, we explore whether the oft-repeated hypothesis of McCarroll et al. (2002), i.e., that BI parameters are more or less representative of lignin content, is relevant for dendroclimatic studies.

2. Materials and methods

The sample material used in this study consists of several datasets of microdensitometric measurements of Pinus sylvestris (Scots pine) trees from the cool and moist boreal forest zone close to the latitudinal tree line in northeastern Finland (200 m a.s.l., Lat 68.9 ^◦N Lon 28.2 ^◦E), sampled in 2014. The data were retrieved from the supplementary material of Bjorklund et al. (2019). From that study, we used an ¨ ensemble of five wood microdensitometric datasets derived using the X-ray based Walesch Electronic Dendro2003 technique (Eschbach et al., 1995). Moreover, we used an ensemble of nine BI-based microdensitometric datasets (Campbell et al., 2007), and one dataset of dendroanatomical density. Each dataset (i.e., ensemble member or chronology) consists of MXD, MXBI, or aMXD (MXD from the dendroanatomical technique) data measurements from 29 sampled trees.

Every tree in each chronology is represented by an average of two to three random and unique radii. In this study, we examine the period 1800–2013; the replication early in each record is lower than 29 trees, i.

e., 9 trees in 1800 CE, 22 trees in 1850 CE and 27 trees in 1900 CE. For more information regarding the sample material, see Bjorklund et al. ¨ (2019).

X-ray, BI, and dendroanatomical microdensitometry generally produce data with different spatial resolutions Bj¨orklund et al. (2019). That is, MXD and MXBI are captured at lower resolutions than aMXD. As a result, the resolution of the dendroanatomical technique has to be reduced to match the effective resolutions of the BI and X-ray techniques. In Bj¨orklund et al. (2019), MXD was estimated to have an effective measurement resolution of 50−60 μm; whereas the corresponding MXBI was estimated to have an effective resolution of 80−120 μm. In the supporting material (Figs. S1-S5), we show that the corresponding resolutions are in fact more accurately estimated to 60 and 120 μm, respectively.

2.1. Dendroanatomical analysis

The dendroanatomical data were measured from digital images of thin sections (15 μm) produced with a sledge microtome (G¨artner et al.

(2015)) equipped with Feather N35 disposable blades (Feather Safety Razor Co., Ltd., Osaka, Japan). Sections were stained with safranin-astrablue to increase contrast and permanently fixed with Canada balsam. Images of the sections were then captured with a camera (Canon EOS 650D, Canon Inc., Tokyo, Japan) mounted on a microscope (Olympus BX41, Olympus Corp., Tokyo, Japan) at a resolution of 2.36 pixels/μm. Multiple overlapping images were stitched into an overall composite image of the anatomical sample using the image-stitching software PTGui (New House Internet Services B.V., Rotterdam, NL; von Arx et al., 2016). The software ROXAS v3.1 was used to automatically detect anatomical structures for all tracheid cells (75–100 radial files per ring) and annual ring borders. The output was summarized as different cell anatomical dimensions, such as lumen, outer cell, and cell wall dimensions, the positions of which were cata- logued within each dated tree ring (Prendin et al., 2017; von Arx and Carrer, 2014). The anatomical parameters were used to calculate dendroanatomical density, i.e., the percentage of wall area of each cell in relation to the full cell area, and integrated into density profiles from which the aMXD and ring widths were extracted (Fig. 1).

(3)

2.2. X-ray and BI analysis

The X-ray density data were produced using the DENDRO2003 – Walesch Electronic measurement setup (Eschbach et al., 1995) accord- ing to standard protocols developed at the Swiss Federal Research Institute WSL (Schweingruber, 1988). Five different datasets from five different labs with similar setups were used: WSL, Switzerland; Dresden, Germany; Krasnoyarsk, Russia; Xi’an, China and Mainz, Germany. The specific measurement protocols used by each specific lab can be found in the supporting material of Bj¨orklund et al. (2019). In general, all labs refluxed their samples in a soxhlet apparatus with ethanol to remove extractives, then exposed the samples to a stationary X-ray source in an environmentally controlled room (c. 50 % RH at c. 20 ^◦C). Analogue X-ray negative films were developed for analysis in the DENDRO2003 densitometer.

The blue intensity data were produced using standard protocols ac- cording to either Campbell et al. (2011) or Rydval et al. (2014). Nine different datasets from nine different labs all using commercial flatbed scanners with various resolutions were used. Again, we refer to the supporting material of Bj¨orklund et al. (2019) for the specific measurement protocols used by each lab. Although the scanners were operated at different nominal resolutions, the nominal resolutions stated for retail scanners are often misguiding (see https://www.imageaccess.

de/_WhitePapers/PDF/WhitePaper_The_Resolution_Myth.pdf, accessed September 2020). For this reason, other means of estimating effective resolution are needed (see supporting information, particularly Figs.

S1-S5). To remove extractives, samples were refluxed using soxhlet extraction with ethanol or placed in a cold bath with acetone. The wood surfaces were prepared using a range of successively finer grit sand paper. Digital images were produced with flatbed scanners that were most often calibrated with SilverFast Ai professional scan software using the calibration target IT8.7/2. The images were analyzed with the commercial software WinDendroTM (Guay et al., 1992) or CDen- dro/CooRecorder (Larsson, 2014).

The rationale for using ensembles of data from different labs is that they enable the quantification of a range of measurements representing the expected variation of values per technique. In other words, ensemble ranges provide a rudimentary confidence interval. If data compared with the ensemble reside outside the expected range, it is likely that the difference is related to the measurement technique. In this case, the difference is related to how relevant the cell wall information is to the xylem density or BI measurement.

2.3. Statistical analyses

First, we correlated raw averaged MXD and MXBI chronologies with corresponding raw averaged aMXD chronologies at matching measurement resolutions. We visually inspected differences among the chronologies to understand whether there were important features that could be attributed to cell wall density or cell wall BI. The major theo- retical difference between aMXD and MXBI and MXD is that only MXBI and MXD integrate information from the cell wall in their measurements. It thus follows that cell wall density and cell wall BI chronologies can be calculated as the arithmetic difference between aMXD and MXD, and aMXD and MXBI, respectively (sensu Decoux et al., 2004; Silkin and Ekimova, 2012; Silkin and Kirdyanov, 2003). The minuends and sub- trahends in these operations were transformed to z-scores prior to sub- traction in order to align their means and variances. We correlated these cell wall chronologies with temperature and precipitation to understand whether the cell wall information in the latewood contains any important climate signals. The cell wall chronologies and climate data were both detrended with cubic smoothing splines prior to correlation analysis (50 % frequency response cut-off at 20 years) (Cook and Peters, 1981). The climate data consist of temperature data retrieved from CRUTEM4 (5^◦gridded monthly dataset) (Osborn and Jones, 2014) and precipitation data from CRU TS 4.03 (0.5^◦ gridded monthly dataset) (Harris et al., 2014). The grid point centered over the sampling site (Lat 65−70 ^◦N, Lon 25−30 ^◦E) comprise data overlapping with the tree-ring Fig. 1. Differences in wood density data acquisition among techniques, with a focus on wood anatomical density (aD). a) X-ray and BI techniques most often use a photo sensor parallel to the ring border (illustrated by the dashed red rectangle) that moves across an X-ray or BI tree-ring image. The sensor builds up a measurement profile of light intensity that is calibrated into wood density or directly used for BI (Bj¨orklund et al., 2019). aD is obtained via a flexible band parallel to the true ring border that moves along a stained micro-section of wood. In a) the flexible band is represented by the area between two blue vertical curves. Cells with lumina that are more than 50 % enclosed by this band (as indicated by yellow-filled lumina for one vertical strip) are included in the calculation of density at that specific position. Here, the radial extension between two blue lines is 60 μm, corresponding to the effective resolution of the X-ray technique. The flexible line is moved at 10-μm steps indicated by the blue double arrows in a). In contrast, the X-ray photo sensor is 10 μm in radial extension (cf. red vertical line) but is also moved in 10 μm steps. The fact that X-ray images can be unfocused and that the photosensor is inflexible can explain why the effective resolution of the X-ray technique is estimated to 60 μm rather than its nominal resolution of 10 μm. b) aD is defined as the proportion of cell wall area (CWA) in relation to the full tracheid area (TA) (i.e., the ratio of the CWA to TA, or CWA/TA). The TA is the area sum of the CWA and the lumen area (LA). A calculation of CWA/TA is performed for each cell, and a median of all anatomical cell densities is calculated for the full length of the flexible band. The successive radial movements of the band build up a measurement profile, illustrated in c). Panel (c) also depicts a so-called “blue ring” in the year 1902. The latewood cell walls in the double-stained microsection have turned blue instead of the normal red because of their low lignin content. This phenomenon occurs almost exclusively in the latewood (LW). The white dots on the profile indicate the maximum latewood density (aMXD) in each year. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

(4)

data in the years 1876–2013 for temperature and 1901–2013 for precipitation.

Finally, we visually compared years, identified with the anatomical images, of so-called “blue rings” (Piermattei et al., 2015). Double staining of micro-sections with safranin and astrablue solutions results in a coloration of the cell walls where lignified cell walls appear red and cell walls with a low lignin content appear blue (Gerlach, 1969;

Schweingruber, 2007). We identified, dated, and summed the number of

“blue rings” for each year. New normalized raw MXD, MXBI, and aMXD chronologies were built based only on trees exhibiting “blue rings” in the particular year of interest. For example, 1805 CE exhibits the blue ring feature in 7 of the 9 trees; thus, new chronologies were built from these 7 trees for MXD, MXBI, and aMXD. We then visually compared 30-year segments centered on the "blue-ring" year. This operation was repeated for all years in which the “blue ring” feature was prominent.

3. Results and discussion

The visual inspection of the MXBI and aMXD chronologies revealed that the two data types are very similar (Fig. 2a). The aMXD (120 μm) is mostly enveloped by the range of MXBI chronologies. However, there are a few exceptions. Just before the turn of the 19th century, the aMXD chronology displays slightly lower values for a decade. Just after the turn of the 19th century, the aMXD chronology displays slightly higher values for several years. The average correlation between the aMXD and the nine MXBI ensemble members is r =0.94. When a 20-year high-pass filter is applied, the average correlation is even higher, at r =0.96 (Fig.

S1). Although the difference between MXBI and aMXD is very small, there appears to be some room for the BI of the cell wall to influence the resulting data. However, this effect appears to be more related to local mean levels than to distinct inter-annual variability. These local mean differences could be related to the subtle difference in reflection that

occurs between the heartwood and the sapwood around this time in most trees.

The X-ray based MXD data and the aMXD (60 μm) data are also very similar (Fig. 2b). In fact, there are no periods in which the aMXD deviates conspicuously from the range of MXD data sets. The average correlation between aMXD and the five MXD ensemble members is r = 0.97. When a 20-year high-pass filter is applied, the average correlation increases further to r =0.98 (Fig. S1). In this case, it is hard to find instances in which cell wall density could have had any substantial influence on the MXD data, especially considering the very narrow range of the MXD ensemble. The potential inter-annual variability of cell wall density must be very limited to obtain these results.

The cell wall BI ensemble, i.e., each MXBI dataset minus the aMXD (120 μm), is highly inter-correlated, with an average of r =0.79. The cell wall density ensemble, i.e., each MXD dataset minus the aMXD (60 μm), is similarly highly inter-correlated, with an average of r =0.78. Stan- dard deviations of the ensemble average cell wall BI and cell wall density chronologies are only 0.26 and 0.22, respectively, compared to standard deviations of 1.0 for both MXBI and MXD. For reference, the ensembles of cell wall BI and cell wall density chronologies can be found in Figure S8. Assuming that the resolution matching was successful, this means that cell wall density and cell wall BI ensemble members contain consistent variability that is distinct from that of MXD. Moreover, the ensemble average of cell wall density chronologies and of cell wall BI chronologies are also highly correlated at r =0.64. This indicates that the cell wall data obtained using the two techniques are very similar.

Given that lignin is the last structural component to be incorporated into the cell wall during formation (Pereira et al., 2003), it is likely that lignin is responsible for any correlated changes in cell wall density or cell wall BI. Increased packing density of lignin may both increase the density of the cell wall and make the cell wall appear darker, thereby explaining the high correlation between cell wall density and cell wall BI.

Fig. 2. Comparison between aMXD chronologies with the corresponding ensembles of MXBI (a) and MXD (b) chronologies. a) The range of the MXBI ensemble is represented by the blue-shaded area; the aMXD (120 μm) chronology is represented by a solid black line. b) Same as in a) but for the MXD ensemble compared with the aMXD (60 μm). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

(5)

In Fig. 3 we compare the monthly climate correlations obtained from ensemble averages of MXBI, MXD, and ring width, as well as aMXD (120 μm and 60 μm), with the correlations obtained for ensemble averages of cell wall density and cell wall BI. MXBI and MXD exhibit significant correlations with temperature from April to August/September, and ring width correlates with July temperature. These correlations with temperature demonstrate that the trees are sensitive to temperature. How- ever, the cell wall chronologies display no significant relationships with monthly temperatures. There are some scattered significant negative correlations between precipitation and ring width, MXD, and MXBI, respectively, but no correlation between precipitation and the cell wall chronologies. It is somewhat surprising that neither cell wall density nor cell wall BI displayed any sensitivity to temperature. Both Piermattei et al. (2015) and Gindl et al. (2000) have hypothesized a connection between the variation in lignin content and late growing season temperatures. It would be reasonable to assume that unfavorable climatic conditions are associated with reduced lignification because the incor- poration of lignin requires extensive photo assimilation. Although cell wall density and cell wall BI show distinct variations, they do not appear to be driven by temperature or moisture.

“Blue rings” (Piermattei et al., 2015) are prominent examples of extreme years in which lignification has partly failed in the latewood.

We know from the meteorological records that the two coldest summers during the period 1876–2013 were 1877 and 1902 CE, which corre- spond to the two distinct “blue ring” years observed in this period (Fig. 4a). The lack of “blue rings” later in the record may be related to the subsequent warming trend and/or to that more mature trees may be less susceptible to the stress of cold temperatures. Thus, even though cell wall BI does not seem to display any significant temperature sensitivity, it appears that lignification fails when growing season temperatures are extremely cold, as also suggested by Piermattei et al. (2015). Gindl et al.

(2000) proposed that the lignin content of the secondary cell wall is sensitive to late growing season temperature. However, our results do not support the theory that lignification fails due to clearly defined late season cold spells (see Figs. 4b and c). But this may simply be related to that cold spells are not detectable in monthly-scale temperature data;

unfortunately, daily temperature records stretching back to 1902 CE are to our knowledge unavailable.

The comparison of individual years exhibiting the “blue ring” phenomenon exhibit no tangible difference between the range of MXBI chronologies and the aMXD (120 μm) chronology. The aMXD chronology is consistently within the range of the ensemble of MXBI

chronologies in the “blue ring” years (Fig. 5a). However, the aMXD chronology deviates somewhat from the MXBI range in several years without “blue ring” features. The most conspicuous difference between the ensemble of the original MXBI chronologies and the aMXD chronology occurs just after 1900 CE (Fig. 2a); surprisingly, this difference disappears in the ensemble of the new subsampled chronologies that includes only trees with “blue ring” years.

The ensemble range of MXD chronologies based on “blue ring” trees exhibits fluctuations that are very similar to those of the aMXD (60 μm) chronology. Although the aMXD chronology is sometimes outside the envelope of the ensemble of MXD chronologies, the offset is persistent Fig. 3.Climate correlations of cell wall BI and cell wall density compared to ring width (RW), MXBI, MXD, and aMXD. a) Correlations between climate and the ensemble average of cell wall BI contrasted against climate correlations with ring width, MXBI, and aMXD (120 μm).

White dots represent statistically significant correlations (α <0.05). b) Correlations between climate and the ensemble average of cell wall density contrasted against climate correlations with ring width, MXD, and aMXD (60 μm). The temperature data were retrieved from the CRUTEM4 5^◦gridded monthly dataset (Osborn and Jones, 2014). The precipitation data are from the CRU TS 4.03 0.5^◦ gridded monthly dataset (Harris et al., 2014). The grid point centered over the sampling site (Lat 65-70 ^◦N, Lon 25-30 ^◦E) comprises data overlapping with the tree-ring data in the years 1876-2013 for temperature and 1901-2013 for precipitation.

All chronologies and climate data were detrended with cubic smoothing splines (50 % frequency response cut-off at 20 years; Cook and Peters, 1981) prior to analysis.

Fig. 4. a) Warm season temperature April-September (AMJJAS) in northern Finland from the CRUTEM4 5^◦gridded monthly dataset (Osborn and Jones, 2014) covering the period 1876-2013 CE. Blue vertical lines are inset on the x-axis to represent all distinct “blue rings” found in the tree-ring data covering 1800-2013 CE. 1877 and 1902 are the years with “blue rings” for which there is meteorological data overlap. In b) and c), the monthly temperatures of 1877 and 1902 CE are presented together with the mean monthly temperatures of 1901-2018 CE. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

(6)

and not related to any pronounced deviation in the “blue-ring” years (Fig. 5b). Even though the “blue rings” are demonstrably less lignified, the importance of low lignin content for cell wall density or cell wall BI appears to be irrelevant for the xylem density and BI of the latewood.

That is, a lower lignin ratio has no significant influence on MXD and MXBI. This may be a result of cell wall shrinkage. If the shrinkage completely diminishes the voids in which the lignin molecules were supposed to go, the density of the cell wall may not necessarily decline, because cellulose and hemi-cellulose are even denser than lignin.

4. Concluding remarks

In this study, we conclude that the inter-annual variability of cell wall composition is irrelevant to the inter-annual variability of xylem density. Moreover, we show that cell wall density and cell wall BI do not appear to contain any climate signals that are in addition to or distinct from those that can already be extracted from the MXD and MXBI parameters. Even in years when the lignification process has been demonstrably reduced by harsh climatic conditions, as indicated by

“blue rings”, we find no consistent influence on the resulting xylem density or BI.

Our results thus reinforce the notion that the xylem density of conifers can be obtained without detectable bias using dendroanatomical techniques, which do not consider variations in cell wall density (cf.

Bj¨orklund et al., 2020). Conversely, techniques that integrate cell wall density and cell wall BI in xylem density estimations do not exhibit improved performance in dendroclimatic studies. Because lignin content appears to play a negligible role, the cell wall BI likely presents the biggest challenge when producing unbiased MXBI data. This is because cell wall BI is notoriously distorted by microbial staining, tree-specific extractive discoloration (heartwood/sapwood; Bj¨orklund et al., 2019;

Wilson et al., 2017), and as a result of conditions in the deadwood preservation environment (Bj¨orklund et al., 2014, 2019; Rydval, Loader et al., 2017; Wang et al., 2020; Wilson et al., 2004). Considering these results, we recommend that future research refrain from repeating the widely cited, though unsubstantiated, hypothesis that the BI technique spectrophotometrically measures lignin content. Unless new evidence is put forth to support this, our results indicate that this hypothesis is misleading.

Although cell wall density and cell wall BI appear to have no potential as new proxy parameters in paleoclimate studies, we emphasize that our quantifications of cell wall density and cell wall BI are integrated on a scale far greater than the individual cell wall. To quantify cell wall density and cell wall composition, very specialized equipment is needed. Analyzing each individual cell wall to create centuries-long chronologies of cell wall information from dozens of trees is currently not feasible. Given the tremendous effort this would take, our results indicate that further research in this direction may be less than fruitful.

Declaration of Competing Interest We declare no conflict of interests.

Acknowledgements

We sincerely thank two anonymous referees for their valuable cri- tiques of an earlier version of the manuscript. This work has been funded by a grant from the Swiss National Science Foundation (Project XELL- CLIM no. 200021_182398) awarded to G.v.A. G.v.A. was also supported by a grant from the Swiss State Secretariat for Education, Research, and Innovation SERI (SBFI C14.0104). P.F. was supported by the Swiss Na- tional Science Foundation (grant no. 183571, CALDERA).

Fig. 5. MXBI, MXD, and aMXD chronologies based on the subset of trees showing prominent

“blue rings” in the [six] years 1805, 1810, 1812, 1857, 1877, and 1902 CE. a) Chronolo- gies from the MXBI ensemble and aMXD (120 μm). b) Same as in a) but for the ensemble of MXD chronologies and aMXD (60 μm). All chronologies are normalized raw chronologies based only on trees exhibiting “blue rings” in the particular year of interest (z-scores were calculated over the period 1800-2013 CE). The blue vertical line indicates the exact year of the blue ring. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

(7)

Appendix A. Supplementary data

Supplementary material related to this article can be found, in the online version, at doi:https://doi.org/10.1016/j.dendro.2020.125785.

References

Bj¨orklund, J., Gunnarson, B.E., Seftigen, K., Esper, J., 2014. Blue intensity and density from northern Fennoscandian tree rings, exploring the potential to improve summer temperature reconstructions with earlywood information. Clim. Past 10 (2), 877.

Bj¨orklund, J., Seftigen, K., Fonti, P., Nievergelt, D., von Arx, G., 2020. Dendroclimatic potential of dendroanatomy in temperature-sensitive Pinus sylvestris.

Dendrochronologia 60, 125673.

Bj¨orklund, J., von Arx, G., Nievergelt, D., Wilson, R., Van den Bulcke, J., Günther, B., et al., 2019. Scientific merits and analytical challenges of tree-ring densitometry.

Rev. Geophys. 57, 1224–1264. https://doi.org/10.1029/2019RG000642.

Brunow, G., Sivonen, M., 1975. Yellowing of lignin. Part II. Participation of oxygen in the photo-dehydrogenation of lignin model compounds. Paperi ja puu 57 (4), 215–216.

Butterfield, B.G., 2003. Wood anatomy in relation to wood quality. In: Barnett, J., Jeronimidis, G. (Eds.), Wood Quality and Its Biological Basis. Blackwell publishing Ltd, Oxford UK, pp. 30–52, 1-84127-319-8.

Campbell, R., McCarroll, D., Loader, N.J., Grudd, H., Robertson, I., Jalkanen, R., 2007.

Blue intensity in Pinus sylvestris tree-rings: developing a new palaeoclimate proxy.

Holocene 17 (6), 821–828.

Campbell, R., McCarroll, D., Robertson, I., Loader, N.J., Grudd, H., Gunnarson, B., 2011.

Blue intensity in Pinus sylvestris tree rings: a manual for a new palaeoclimate proxy.

Tree. Res. 67 (2), 127–135.

Cook, E.R., Peters, K., 1981. The smoothing spline: a new approach to standardizing forest interior tree-ring width series for dendroclimatic studies. Tree-Ring Bulletin.

Decoux, V., Varcin, ´E., Leban, J.M., 2004. Relationships between the intra-ring wood density assessed by X-ray densitometry and optical anatomical measurements in conifers. Consequences for the cell wall apparent density determination. Ann. For.

Sci. 61 (3), 251–262.

Elliott, G.K., Brook, S.E.G., 1967. Microphotometric technique for growth-ring analysis.

J. Inst. Wood Sci. 18, 24.

Eschbach, W., Nogler, P., Sch¨ar, E., Schweingruber, F.H., 1995. Technical advances in the radiodensitometrical determination of wood density. Dendrochronologia 13, 155–168.

G¨artner, H., Lucchinetti, S., Schweingruber, F.H., 2015. A new sledge microtome to combine wood anatomy and tree-ring ecology. IAWA J. 36 (4), 452–459.

Gerlach, D., 1969. Botanische mikrotechnik. Georg Thieme Verlag.

Gindl, W., Grabner, M., Wimmer, R., 2000. The influence of temperature on latewood lignin content in treeline Norway spruce compared with maximum density and ring width. Trees 14 (7), 409–414.

Guay, R., Gagnon, R., Morin, H., 1992. A new automatic and interactive tree ring measurement system based on a line scan camera. For. Chron. 68 (1), 138–141.

Harris, I.P.D.J., Jones, P.D., Osborn, T.J., Lister, D.H., 2014. Updated high-resolution grids of monthly climatic observations–the CRU TS3. 10 Dataset. Int. J. Climatol. 34 (3), 623–642.

Hill, C.A., Papadopoulos, A.N., 2001. A review of methods used to determine the size of the cell wall microvoids of wood. J. Inst. wood Sci. 15 (6; ISSU 90), 337–345.

Jacquin, P., Longuetaud, F., Leban, J.-M., Mothe, F., 2017. X-ray microdensitometry of wood: A review of existing principles and devices. Dendrochronologia 42, 42–50.

Jagels, R., Telewski, F.W., 1990. Computer-aided image analysis of tree rings. Methods Dendrochronol. 76–93.

Larsson, L., 2014. CooRecorder And Cdendro Programs of the CooRecorder/Cdendro Package Version 7.7..

Leary, G.J., 1968. The yellowing of wood by light: part II. Tappi 51 (6), 257–260.

Ljungqvist, F.C., Thejll, P., Bjorklund, J., Gunnarson, B.E., Piermattei, A., Rydval, M., ¨ Seftigen, K., Stove, B., Buntgen, U., 2020. Assessing non-linearity in European temperature-sensitive tree-ring data. Dendrochronologia 59, 125652.

McCarroll, D., Pettigrew, E., Luckman, A., Guibal, F., Edouard, J.L., 2002. Blue reflectance provides a surrogate for latewood density of high-latitude pine tree rings.

Arct. Antarct. Alp. Res. 34 (4), 450–453.

Osborn, T.J., Jones, P., 2014. The CRUTEM4 land-surface air temperature data set:

construction, previous versions and dissemination via Google Earth. Earth Syst. Sci.

Data Discuss. 6 (1), 61–68.

Pereira, H., Graça, J., Rodrigues, J.C., 2003. Wood chemistry in relation to quality. In:

Barnett, J., Jeronimidis, G. (Eds.), Wood Quality and Its Biological Basis. Blackwell publishing Ltd, Oxford UK, pp. 53–86, 1-84127-319-8.

Piermattei, A., Crivellaro, A., Carrer, M., Urbinati, C., 2015. The “blue ring”: anatomy and formation hypothesis of a new tree-ring anomaly in conifers. Trees 29 (2), 613–620.

Polge, H., 1978. Fifteen years of wood radiation densitometry. Wood Sci. Technol. 12 (3), 187–196.

Prendin, A.L., Petit, G., Carrer, M., Fonti, P., Bj¨orklund, J., von Arx, G., 2017. New research perspectives from a novel approach to quantify tracheid wall thickness. Tree Physiol. 37 (7), 976–983.

Rydval, M., Larsson, L.Å., McGlynn, L., Gunnarson, B.E., Loader, N.J., Young, G.H., Wilson, R., 2014. Blue intensity for dendroclimatology: should we have the blues?

Experiments from Scotland. Dendrochronologia 32 (3), 191–204.

Rydval, M., Loader, N.J., Gunnarson, B.E., Druckenbrod, D.L., Linderholm, H.W., Moreton, S.G., Wood, C.V., Wilson, R., 2017. Reconstructing 800 years of summer temperatures in Scotland from tree rings. Clim. Dyn. 49 (9–10), 2951–2974.

Saranp¨a¨a, P., 2003. Wood density and growth. In: Barnett, J., Jeronimidis, G. (Eds.), Wood Quality and Its Biological Basis. Blackwell publishing Ltd, Oxford UK, pp. 87–117, 1-84127-319-8.

Schweingruber, F.H., 1988. Tree Rings: Basics and Applications of Dendrochronology.

Kluwer Academic Publishers, Dordrecht, Netherlands; Boston, Massachusetts, USA, p. 276.

Schweingruber, F.H., 2007. Wood Structure and Environment. Springer Science &

Business Media.

Silkin, P.P., Ekimova, N.V., 2012. Relationship of strontium and calcium concentrations with the parameters of cell structure in Siberian spruce and fir tree-rings.

Dendrochronologia 30 (2), 189–194.

Silkin, P.P., Kirdyanov, A.V., 2003. The relationship between variability of cell wall mass of earlywood and latewood tracheids in larch tree-rings, the rate of tree-ring growth and climatic changes. Holzforschung 57 (1), 1–7.

Stamm, A.J., Sanders, H.T., 1966. Specific gravity of wood substance of loblolly pine as affected by chemical composition. Tappi 49 (9), 397.

Tsoumis, G., 1964. Microscopic measurement of the amount of cell wall substance in wood and its relationship to specific gravity. Tappi 47 (11), 675–677.

Vaganov, E.A., Hughes, M.K., Shashkin, A.V., 2006. Growth Dynamics of Conifer Tree Rings: Images of Past and Future Environments Vol. 183. Springer Science &

Business Media.

von Arx, G., Carrer, M., 2014. ROXAS–A new tool to build centuries-long tracheid-lumen chronologies in conifers. Dendrochronologia 32 (3), 290–293.

von Arx, G., Crivellaro, A., Prendin, A.L., Čufar, K., Carrer, M., 2016. Quantitative wood anatomy—practical guidelines. Front. Plant Sci. 7, 781.

Wang, F., Arseneault, D., Boucher, E., Yu, S., Ouellet, S., Chaillou, G., Wang, L., 2020. ´ Chemical destaining and the delta correction for blue intensity measurements of stained lake subfossil trees. Biogeosciences 17 (18), 4559–4570.

Wilfong, J.G., 1966. Specific gravity of wood substance. For. Prod. J 16 (1), 55–61.

Wilson, R.J., Esper, J., Luckman, B.H., 2004. Utilising historical tree-ring data for dendroclimatology: a case study from the Bavarian Forest, Germany.

Dendrochronologia 21 (2), 53–68.

Wilson, R., D’Arrigo, R., Andreu-Hayles, L., Oelkers, R., Wiles, G., Anchukaitis, K., Davi, N., 2017. Experiments based on blue intensity for reconstructing North Pacific temperatures along the Gulf of Alaska. Clim. Past.

Yiannos, P.N., 1964. The apparent cell-wall density of wood and pulp fibers. Tappi 47 (8), 468.