Seasonal patterns of bole water content in old growth Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco)

Abstract

Large conifer trees in the Pacific Northwest, USA (PNW) use stored water to extend photosynthesis, both diurnally and seasonally. This is particularly important during the summer drought, which is characteristic of the region. In the PNW, climate change is predicted to result in hotter, drier summers and warmer, wetter winters with decreased snowpack by mid-century. Understanding seasonal bole water dynamics in relation to climate factors will enhance our ability to determine the vulnerability of forests to climate change. Seasonal patterns of bole water content in old-growth Douglas-fir (Pseudotsuga menziesii (Mirb.) Franco) trees were studied in the Cascade Mountains of western Oregon, USA. Relative water content (RWC) was monitored hourly in three 400+ and three ~150 years-old trees using permanently mounted dielectric devices for 10 years. RWC increased during the late spring and early summer to maximum levels in August then decreased into fall and remained low over winter. The difference between minimum RWC in the winter and maximum in mid-summer averaged 4.5 and 2.3% for the older and younger trees, respectively, across all years. RWC closely followed growth and was positively correlated with air and soil temperature, vapor pressure deficit and photosynthetically active radiation, but lagged plant available soil water. The progressive decrease in RWC seen each year from mid-summer through fall was attributed to net daily loss of water during the summer drought. The marked increase in RWC observed from spring to mid-summer each year was hypothesized to be the period of embolism repair and water recharge in elastic tissues. We conclude that bole water content is an integral part of tree water dynamics enabling trees to extend carbon assimilation into drought periods and during periods when cold soil inhibits water uptake by roots, an adaptation that could benefit the survival of large PNW trees under climate change.

1. Introduction

In the Pacific Northwest (PNW), global climate change is predicted to result in hotter, drier summers and warmer winters with substantial decreases in snowpack by mid-century (Mote and Salathé, 2010). Over the past century the PNW has experienced increased annual mean temperature of 0.68–0.88 °C with warming in most seasons, and observations show that decreased summer and autumn precipitation has resulted in increased potential evapotranspiration resulting in larger climatic water deficits (Abatzoglou et al. 2014). Increases in the frequency, duration, and severity of drought and heat stress associated with climate change could fundamentally alter the structure and functioning of forest ecosystems (Allen et al., 2015).

The ability of large conifer trees in the PNW to use stored water to extend photosynthesis, both diurnally and seasonally, is thought to be an adaptation to seasonal drought, which typically occurs from July to October each year (Waring and Franklin 1979). Douglas-fir (Pseudotsuga menziesii (Mirb.) is a major component of PNW forests and is the major timber species in the region. Water stored in the boles of old-growth Douglas-fir trees is important in supporting daily and seasonal transpiration (Ćermák et al. 2007, Phillips et al. 2003, Waring and Running 1978). Based on sap flow measurements, large Douglas-fir trees use stored water during the daytime when transpiration exceeds root uptake and recharge during times of low transpiration resulting in little to no net diel change provided soil water is not limiting (Ćermák et al. 2007). Stored water can account for over 20% of daily sap flow allowing an estimated 18% more photosynthesis than if stored water was not used (Phillips et al. 2003). The use of stored water increases throughout the summer drought, supplementing transpirational loss by a progressive decrease the net daily amount of bole water (Ćermák et al. 2007). Waring and Running (1978) estimate that sapwood water content can decrease by as much as 5% per day during periods of high evaporative demand, which in large trees could represent 7% of daily water use (Phillips et al. 2003).

Bole water includes that present in the sapwood as well as the bark, including cambium, phloem and cortex. Water content of the two compartments, although interdependent, change by different processes (Zweifel et al. 2001, Ćermák et al. 2007); water loss from the sapwood results from cavitation, while elastic and reversible shrinkage of the living cells is due to water movement from the bark to the sapwood. Vertical movement in the xylem follows a water potential gradient from soil to roots to stem to branches to the atmosphere and is regulated by the stomata (Zweifel et al. 2007). Movement between the extensible tissues of the living bark and xylem also follows a water potential gradient wherein movement results in shrinkage and swelling of the stem driven by changing water potential in the xylem. This shrink-swell has been related to the water flow dynamics of trees (Steppe et al. 2005), and implies that changes in stem diameter should be directly related to bole water content in the absence of growth.

A better understanding of seasonal use of stored water in trees and the associated climate drivers will enhance our ability to identify species and habitats most vulnerable to climate change. Unfortunately, information on seasonal dynamics of bole water content spanning more than a growing season is scarce.

Waring and Running (1978) found that the sapwood water content in old-growth Douglas-fir decreased starting in the spring and continuing until the fall by as much as 50%; it then increased through the winter. In addition to old-growth Douglas-fir (Waring and Running 1978), progressive, summer decreases in bole water content corresponding to drought conditions was seen in sapwood of mature ponderosa pine (Pinus ponderosa Dougl.) where the decrease was substantially less than in Douglas-fir (Domec et al. 2005). Waring et al. (1979) reported a 27% decrease bole water content over a fortnight in Scots pine (Pinus sylvestris L.) sapwood. Domec and Gartner (2001, 2002a) found decreases in sapwood water content of up to 20% in both mature and sapling Douglas-fir stems from winter to late summer. In these studies, tissue samples were collected with increment borers, which can result in sample drying (Constantz and Murphy 1990, Hao et al 2013, Morales et al. 2001).

Instruments that estimate water content of soil, wood and other substances using changes in apparent dielectric constant, have been used for in situ measures of water in tree boles (Constantz and Murphy 1990). Such devices allow continuous data collection from a single position on the bole with minimal disturbance to tissues. A four-year study of two 110-120 years old Douglas-fir trees using permanently mounted moisture probes showed increasing bole water content from the onset of growth in the spring until mid-summer when it decreases to a relatively consistent level over winter (Beedlow et al. 2007b). Further, that study showed that bole water fluctuates only ~5% annually. Similar results have been shown using moisture probes permanently placed into hardwood species and other conifer species (Irvine and Grace 1997, Irvine et al. 1998, Wullschleger et al. 1996). Beedlow et al. (2007b) hypothesized that the discrepancy between their findings and those of Waring and Running (1978) in Douglas-fir may be the result of differences in tree size or age. Tree size and age can affect wood characteristics and, consequently, water storage capacity (Domec and Gartner 2001, 2002a, Phillips et al. 2003). Waring and Running (1978) sampled trees 400+ years old, whereas Beedlow et al. (2007b) studied smaller trees ~115 years old.

To test the hypothesis that tree size and age affect seasonal patterns of bole water content in Douglas fir, we equipped six Douglas fir trees located in a mixed conifer stand with commercially available dielectric devices mounted into the boles. Three of the study trees were ~150 years old, and three were 400+ years old, and ranging in diameter from approximately one half to two meters. We measured bole water content on an hourly basis from November 2004 through December 2014 in conjunction with soil water, and meteorological conditions in order to examine the seasonal patterns of bole waters and potential environmental factors driving those patterns.

2. Materials and methods

Two groups of three trees were selected at an established monitoring site located approximately 90 km southeast of Salem, Oregon, USA on the west slope of the Cascade Mountains (44° 21′ N, 122° 17′ W). The site was on glacial deposits at 1200 m elevation. Air temperature averaged 1.7° C in January, and 23.8 °C during July and August over the study period, November 2004 through December 2014. Mean annual precipitation was 2014 mm over that time, averaging 213 mm from July through September—the seasonal drought period. Douglas-fir and western hemlock (Tsuga heterophylla (Raf.) Sarg.) dominated the stand with silver fir (Abies amabilis (Dougl. ex Loud.) Dougl.) reaching the canopy and an understory of widely scattered vine maple (Acer circinatum Pursh). The younger trees, 140, 152 and 155 years old, were located on the western edge of the site, and the older trees, 400+ years, were located along the northern edge (Table 1). The pith date of the 400+ years old trees could not be determined because our increment borers were unable to reach the center of the trees, but other Douglas-fir in the stand that were successfully cored exceeded 520 years. Although heart rot is common in old growth Douglas-fir, the study trees showed no external signs of disease or physical damage, and the outer 250 mm of bole wood was solid.

Table 1.

Characteristics of the study trees. Tree numbers are shown with age class identifiers in parentheses; O=older, 400 years+, Y=younger, 140–155 years. Diameter at breast height (DBH) measured at approximately 1.4 m; basal area (BA). The cortex or outer bark was divided into outer and inner portions (OB′), which was the portion between the base of the ThetaProbe body and the phloem (PH); sapwood (SW) and heartwood (HW).

	Tree
	1(O)	2(O)	3(O)	4(Y)	5(Y)	6(Y)
Ht (m)	59	54	54	40	41	43
Age (yrs)	400+	400+	400+	152	140	155
2004 DBH (m)	1.94	1.28	1.64	0.74	0.67	0.81
BA 2004 (m2)	2.96	1.28	2.10	0.43	0.35	0.51
BA increase 2004–14 (%)	3.5	7.2	2.7	9.4	8.6	14.5
OB′ growth (mm) 2004–14	21.68	23.07	5.64	0.66	1.66	6.04
PH growth (mm) 2004–14	3.92	6.46	3.30	3.96	3.29	4.05
SW growth (mm) 2004–14	14.72	29.76	12.24	14.73	14.30	20.60

CR10X or CR1000 dataloggers (Campbell Scientific, Inc., USA) were used to record numerous above- and belowground parameters within 10 m of each group of three trees at 5-minute intervals and saved as hourly averages. Air temperature and relative humidity (RH) were measured 3m above the ground and at the top of the forest canopy using Campbell Scientific HMP45 temperature–humidity sensors. Air temperature at the boles (Ta) was measured at ~1.5 m on the north side of the trees (Campbell Scientific 107 Thermistors). Precipitation (PPT) was measured and summed hourly at the canopy top using a Texas Electronics TE525I tipping bucket. Photosynthetically active radiation (PAR) was measured at the top of the canopy using LI-COR, LI-190SL sensors (Lincoln, NE, USA). Vapor pressure deficit (VPD, MPa) was calculated from air temperature and RH data taken at the canopy top. Soil temperature (Ts) was measured at depths of 0.05, 0.15, and 0.30 m (Campbell Scientific 107 Thermistors). Volumetric soil moisture (Campbell Scientific CS-615 reflectometers) was measured over 0.2 m increments to a depth of 0.6 m. Plant available soil water (ASW, mm) was calculated for the top 0.6 m of soil from the volumetric measures and moisture release curves developed for each site. Beedlow et al. (2013) provide more detailed stand characteristics as well as weather and soil monitoring instrumentation for this site, named “Soapgrass.”

2.1. Bole water measurement

ThetaProbe moisture sensors (Delta-T Devices Ltd, Cambridge, UK) measured bole water content at the same time interval as the weather and soil data. The device measures volumetric water content (±1% v/v, from 0–40 °C) by the well-established method of measuring changes in the apparent dielectric constant, but it uses a capacitance, rather than a time-domain, approach (Gaskin and Miller 1996). Changes in the apparent dielectric constant are directly proportional to water content. The device measures water content within a cylinder of ~20 mm in radius along the 60 mm length of the probe rods. The ThetaProbes were calibrated to measure percent relative water content (RWC; %) of the bole tissues following the procedure suggested by the manufacturer, and discussed by Gaskin and Miller (1996) on freshly cut bole sections as detailed in Beedlow et al. (2007b). The manufacturer has minimized the influence of temperature on the output of the ThetaProbe by a compensating technique (Gaskin and Miller 1996). We tested the actual sensitivity of the probes over the range of temperatures expected in the field and found the instruments stable (Beedlow et al. 2007b).

Using unpublished data on seasonal changes in sapwood and phloem sugar concentrations in mature Douglas-fir, we tested the ThetaProbes for the effect of sugar concentration across a range of 0 to 500 micro-mol/g water (similar to maximum concentrations found in Douglas-fir phloem). Increasing concentrations of sugar lowered the apparent RWC from the probes by 0.0046 % per micro-mol/g. This suggested that the effects of seasonal changes in sugar concentrations on RWC would be to lower apparent RWC by ~0.2 % in winter and by ~.01 % in summer. Because the effect would be to decrease the apparent RWC in both winter and summer, we did not adjust the ThetaProbe reading for sugar concentration.

ThetaProbes were installed in the stems, or boles, at the base of each tree at approximately 1.4 m from the ground on the north side. Additional probes were installed in the boles at the base of the canopies (hereafter, “canopy-level”) of the 400+ years old trees in the summer of 2007. The boles were prepared by removing bark with a Forstner bit to within 5 mm of the phloem. Guide holes for the probe rods were drilled 60 mm into the boles using a jig to keep the holes aligned. The diameter of the guide holes was slightly smaller than that of the probe rods to maximize rod-wood contact. Even though we took efforts to insure tight contact between the probe rods and wood, data taken during the first growing season (2005) were erratic and generally decreasing, which indicated that a “grow-in” period is required (Figure 1). ThetaProbes were calibrated to measure the relative water content (RWC) of the bole tissues following the procedure detailed in Beedlow et al. (2007b). RWC is the amount of water in the sample divided by the potential maximum amount of water in the sample assuming a cell wall density of 1530 kg m⁻³ (Domec and Gartner 2002a, Waring and Running 1978).

Figure 1.

Mean daily relative water content (RWC) in the older trees (O) showing annual patterns of bole water fluctuations at A) canopy-level and, B) the base of the trees. Marked decreases in RWC during winter are the result of freezing. Decreasing RWC during the first year of sensor operation is apparent both at canopy-level and at the base and is likely a wound response. C) Percent change per day in relative basal area increment (_rBAI), and D) total plant available soil water (ASW) in the upper 0.6 m of the soil. Shaded areas identify the period 1 June - 1 October, the approximate the growing season. Blanks in the data traces are periods of missing data.

2.2. Tissue measurements

The ThetaProbes provided an integrated measure of RWC for tissues in contact with the 60 mm probe rods. At the start of the study the probe rods penetrated into the sapwood SW of five trees and into the heartwood of one tree. The proportion of the probe rods situated in a given tissue type was measured during probe installation in November 2004 and again before the onset of new growth in April 2015. The length of each tissue type was measured with a digital caliper (Mititoyo, Inc., Japan) accurate to ±0.02 mm. Tissues were identified as cortex or outer bark (OB), phloem (PH), sapwood (SW) and heartwood (HW). For the April 2015 measurement, OB was divided into an outer portion (OB) and an inner portion (OB′) with OB′ as the length between the base of the ThetaProbe body and the PH. During installation (2004) OB and PH lengths were directly measured in the hole drilled through the cortex for probe installation. In April 2015, access holes were drilled with a Forstner bit along the left and right sides of the probe bodies, allowing us to re-measure the OB, OB′ and PH lengths along the probe rods. In both 2004 and 2015 increment bores (5mm, Suuntu, Sweden) taken within ~200 mm of the ThetaProbes were used to estimate the length of SW and HW tissue measured by the ThetaProbe measurement rods. Tissue lengths adjacent to the probes mounted in the canopy were not re-measured in 2015 because of the risk of damaging the small boles.

Differences in water content between tissue types were determined gravimetrically. Evaporative loss from the cores was minimized by collecting samples in November 2008, 2013, and 2014 during cool weather when VPD was low. At each date, each study tree was cored once adjacent to, but >20 mm away from the probes using a 5 mm increment borer (Suuntu, Sweden). Cores were separated by tissue type and the samples immediately placed into pre-weighed vials sealed with airtight lids and transported in an insulated box to the lab for gravimetric determination of water content. The fresh weight of each sample was measured and the samples were oven dried at 60 °C for 5 to 7 days, then reweighed. We used percent water on a fresh weight basis rather than RWC to compare water content between tissues. This was because core lengths are required for RWC computations, and length measures for OB, OB′ and PH were difficult to determine from cores because the tissues expanded or crumbled when removed from the borer.

2.3. Basal Area Increment

Manually read Series 5 band dendrometers (Agricultural Electronics Corp., Tucson AZ; accurate to within a 0.1 mm change in circumference) were installed 0.1 m below the ThetaProbes in November 2004 to measure radial growth. Data were collected biweekly from June through August and at 4-to-6-week intervals the rest of the year throughout the study. During installation, the loose, outermost layers and decaying portions of bark were removed with a rasp so that the dendrometer bands seated firmly against solid bark. To constrain the effect of diurnal tree water use on measures of circumference, readings and samples were taken between 10 AM and 1 PM Pacific Standard Time. The data were adjusted for the effect of temperature on the metal bands by normalizing the field reading to 10 °C. This was done by adding to the field reading, the product of the air temperature at the time of the reading less 10 °C, the band length, and the coefficient of thermal expansion of the band material (Hastelloy C276, 11.2 μm m⁻¹ °C ⁻¹). See Beedlow et al. (2007b and 2013) for more on temperature adjustment to dendrometer data.

Because the study trees differed in size, relative basal area increment (_rBAI; Equation 1) was used to compare growth patterns. _rBAI was expressed as the percent change in basal area per day from any given sample date to the next (% d⁻¹) to account for different numbers of days between sample dates throughout the study period.

$${}_{\mathrm{r}}\mathrm{B}\text{AI}=(({\text{BA}}_{\mathrm{t}+1}-{\text{BA}}_{\mathrm{t}})/(d_{\mathrm{t}+1}-d_{\mathrm{t}}))/{\text{BA}}_{\mathrm{t}}\times 100\%$$ [1]

where: _rBAI = relative basal area increment, BA = measured basal area, d = Julian day, and t = time index.

3. Results

RWC as measured by the ThetaProbes is a function of the dielectric properties of the cell wall material, and the amount of air and water in the various tissues contacting the probe rods. Assuming the dielectric constant of cell walls is consistent (1.4 to 2.9 for dry wood of various species), changes in RWC result from changes in the relative amount of water and air in the tissues. As air has a dielectric constant of ~1 while that of liquid water is ~80, the more air in the tissues the lower the RWC. The amount of air in the tissues can increase as the bark dries and through embolisms in the capillary water within the trachea lumens of the sapwood resulting from water stress (Ćermák et al. 2007). Apparent decreases in tissue RWC can result from ice formation in the tissues. The dielectric constant of ice is ~4 resulting in an apparent drop in moisture as water in the tissue freezes (Sparks et al. 2001).

RWC increased each year from May to early August and then decreased through the summer drought into the fall holding minimum values over winter in the boles at both the canopy bases and (Figure 1A) and the lower boles at the base (Figure 1B) of the older trees, and in the lower boles of the younger trees (Figure 2A). The seasonal patterns of bole water content (RWC) were similar at canopy height and the base of the trees. During the winter months, RWC held relatively level except during periods of freezing when dramatic drops in apparent tissue moisture occurred (Figures 1A, 1B, and 2A). The frequency and magnitude of freezing events—indicated by sharp decreases in RWC during winter—was higher in the boles at canopy-level than at the base of the trees (Figure 1, A and B). Excluding freezing periods and the grow-in years, 2005 for the lower boles and 2007 for the boles in the canopies, the average difference between the maximum and minimum RWC within years was relatively small, ranging from 2.7 to 5.5% in the lower boles and 3.4 to 6.7% in boles at canopy-height across all trees, regardless of age. In both younger and older trees the seasonal pattern of _rBAI matched that of RWC (Figures 1C and 2B), increasing from late spring to mid-summer and declining through the fall and into winter, but _rBAI usually peaked before RWC. _rBAI was roughly two times higher in the younger trees.

Figure 2.

A) Mean daily relative water content (RWC) in the boles of the younger trees (Y) showing annual patterns of bole water fluctuations. Marked decreases in RWC during winter are the result of freezing. The decreasing and somewhat erratic changes in RWC during the first year of sensor operation is likely a wound response. B) Percent change per day in relative basal area increment (_rBAI), and C) total plant available water (ASW) in the upper 0.6 m of the soil. Shaded areas identify the period 1 June – 1 October, the approximate the growing season. Blanks in the data traces are period of missing data.

The annual change in ASW was proportionately much greater than RWC. ASW decreased an average of 81% from winter to mid-summer each year near the older trees and 67% near the younger trees. ASW began decreasing in May or early June and reached minimum levels in late August or September in the upper 0.6 m of the soil in both groups of trees. Both maximum and minimum ASW were less around the older trees (Figure 1D) than the younger trees (Figure 2C). Spikes in ASW during the fall and winter correspond to periods when soil water exceeded field capacity, usually during high rainfall or rapid snowmelt events. These spikes were excluded when calculating yearly maximum ASW over the 10 years.

Bole RWC showed an increasing trend over the study period in the canopies (Figure 1A) and at the base of Trees 1, 2, 3 and 6, but not Trees 4 and 5 (Figures 1B and 2A). Changes in tissue proportions along the probe rods in combination with differences in tissue water content indicated that the trend of increasing RWC resulted from proportional changes in tissue types along the Theta Probe rods. The probes provided an integrated RWC measurement over OB′, PH, all or a portion of the SW and, initially in one tree, a portion of the HW. Because the probe body was mounted outside of the cambium, the entire probe was pushed outward as the trees grew. As a result, tissue proportions along the probe rods changed over the study period. Gravimetric water content varied between tissues among the trees although the differences were not always significant (paired t-test, p<0.1). Briefly, OB′ tended to be wetter than SW except in Tree 5(Y); PH also tended to be wetter than SW except Tree 3(O); and OB′ was similar to PH, but significantly wetter than PH in Tree 3(O) and significantly drier in Tree 5(Y).

Trees with the greatest proportional increase in OB′ had stronger increasing trends in RWC over the study period. Those trees also had the greatest decreases in the proportion of SW. Increasing annual maximum and minimum RWC among trees was directly related to increasing OB′ from 2004 to 2014 (Figure 3A). Similarly, RWC was inversely related to the changing proportion of SW along the probe rods during the same time (Figure 3B). The greatest increase in RWC and OB′ occurred in Trees 1(O) and 2(O), which also had the greatest decrease in SW along the probe rods. Trees 3(O) and 6(Y) showed an intermediate increase in maximum RWC with increases in OB′ of about 10%, and decreases in SW about 15%. Trees 4(Y) and 5(Y) showed virtually no increase in RWC over time and less than a 5% change in OB′ and SW. Based on this information, we de-trended the RWC data for each tree.

Figure 3.

Change in bole relative water content (RWC) compared with, A) changes in the proportion of cortex (OB′), and B) changes in the proportion of sapwood (SW) along the probe rods between the 2004 and 2014 growing seasons. Labels on the data points represent the tree number and age class, older (O) and younger (Y).

Seasonal patterns of bole RWC at canopy-level of the older trees reflected those at the base, but canopy-level RWC peaked an average of 3 days earlier than at the base. Diel patterns of decreasing RWC during the daytime hours and increasing RWC during nighttime hours were not clear in the boles, nor the canopies while RWC was increasing day-to-day during spring and early summer. But as the increase of RWC slowed near the peak, diel fluctuations of RWC on the order of 0.5% became visible each year appearing first in the canopies. Daytime draw-down and nighttime refilling continued through late summer and fall, briefly interrupted by rain evets. As an example, Figure 4 shows these patterns in 2009 averaged for the boles at canopy-level and at the base of the older trees. From 26 June to 24 July 2009, RWC in both locations was increasing day-to-day and diel patterns were inconsistent (Figure 4A). Between 24 July and 4 August 2009, RWC peaked at both heights and showed distinct diel patterns of decreasing RWC from sunrise to early afternoon and increasing RWC from late afternoon until around midnight (Figure 4B). The base typically lagged the canopy-level by 1–3 hours in all phases of this diel cycle. The diel pattern of decreasing and increasing RWC continued through late summer and fall as represented by the period 2–15 September 2009. During this time RWC was generally decreasing at both the base and canopy-level, but it was sensitive to rain events. The effect of rain was to disrupt the diel pattern and result in increasing RWC over a period of days followed by a resumption of day-to-day decreasing RWC (Figure 4C).

Figure 4.

Diel patterns of RWC in the bole at the base and canopy-level of the older trees during early, mid-, and late growing season. A) 26 June to 24 July 2009, RWC was increasing day-to-day and diel patterns were inconsistent. B) 24 July and 4 August 2009, RWC peaked and distinct diel patterns of RWC were visible. C) 2–15 September 2009, the diel patterns continued, and RWC was generally decreasing, but it was sensitive to rain events. RWC was normalized to the 1 May to 31 October, 2009 mean. Alternating shaded areas are used to help distinguish 24 hour periods starting at midnight.

To examine the seasonal patterns of RWC and environmental influences, the de-trended RWC time series data for each tree were averaged for the older and younger trees separately and displayed using a cubic spline smoother with a 50% frequency response of 32 years to remove as much high-frequency variability as possible. The seasonal patterns of RWC compared with _rBAI and ASW is shown in Figure 5. RWC peaked an average of 26 days after _rBAI in the older trees and 36 days in the younger trees. Minimum RWC values occurred in late winter and coincided with low _rBAI over the winter. RWC peaked typically in August before ASW reached minimum levels. Peak RWC in the boles of older trees occurred when ASW in the top 0.6 m of soil reached an average of 44% of maximum, ranging from 37–50% over the study period. Similarly, in the younger trees ASW at peak RWC in the boles averaged 38% of maximum ranging from 35–40%. Peak RWC in the boles at canopy-level in the older trees occurred when ASW reached an average of 44% ranging from 39–51%.

Figure 5.

Seasonal changes in bole RWC, _rBAI and ASW. A) Younger trees, B) Base of the older trees. Time series data are de-trended and smoothed using a cubic spline with a 50% frequency response of 32 years. Shaded areas represent the months of June through September. Each factor was normalized to the overall mean and scaled for ease of comparison. A) RWC=ND(1)+52.6, _rBAI=ND(0.003)+0.004, ASW=ND(25)+227.1; B) RWC=ND(1.689)+47.6, _rBAI=ND(0.001)+0.002, ASW=ND(25)+140.3, where ND=normalized data on the y-axis of each panel.

Pearson correlation analysis of the de-trended RWC data by month indicated that during most of the year, RWC was negatively related to ASW and positively correlated to VPD, Ts, maximum daily Ta (Tmax) and PAR, primary factors affecting sap flow (Figure 6). The correlation in August was highly variable for all parameters as it was consistently a transition month where atmospheric values reached maximum and ASW reached minimum. In August, correlations between parameters varied positive and negative from year to year. The relation of RWC with ASW was quite variable over the winter and early spring, but was negatively correlated with ASW from May through December. In February through April, RWC was positively correlated with ASW as RWC began to increase while ASW was also increasing to an April peak. VPD showed a strong positive correlation with RWC from April through July, but the correlation decreased markedly in the older trees with higher variability in August. VPD showed strong positive correlation in the fall, and negative but variable correlation in the winter. Ts typically started increasing from winter lows of <6 °C in May and showed a strong positive correction with RWC except in February and March when it became negative because Ts was still decreasing while RWC starting increasing. Tmax correlated positively with RWC throughout the year, although the variability increased in August and during the winter. PAR was positively correlated with RWC during spring, but became negative in early to mid-summer as RWC was still increasing but solar radiation began to decrease.

Figure 6.

Seasonal changes in Pearson’s Correlation Coefficients of Relative Water Content (RWC) with environmental factors. ASW is plant available soil water, VPD is vapor pressure deficit calculated from air temperature and relative humidity, Ts is the soil temperature at the 0.2–0.4 m depth increment, Tair is maximum daily air temperature at the bole, and PAR is total photosynthetically active solar radiation. Positive values are positive correlations. Symbols represent the median values. Vertical bars represent the 75^th and 25^th percentiles. The alternating shaded and unshaded areas denote months of the year and are simply an aid for comparing data between panels.

4. Discussion

Our findings do not support the hypothesis that tree size and age affect the seasonality of bole water content. The seasonal patterns in bole RWC in 115–120 years old Douglas-fir (Beedlow et al. 2007b) and the 140–155 and 400+ years old Douglas-fir in this study are similar. In both cases, RWC increased markedly with the onset of growth in the spring, peaked in mid-summer, and then decreased to minimum levels during the fall and winter. Despite slight changes during winter, as well as marked, short-term drops in RWC during freezing events, substantial and consistent increases did not occur until spring. This is in contrast to the pattern of bole RWC reported in old growth Douglas-fir by Waring and Running (1978) where RWC began to increase in the fall from summer lows to maximum levels in the winter before starting to decrease with the onset of growth in the spring. Studies using increment cores to measure RWC reported decreases of up to 50% from winter to summer (Domec and Gartner 2001, 2002a, Waring and Running 1978). Here, the change in bole RWC from winter lows to peak levels during the growing season averaged only 2.7% to 5.5% in older and younger trees, respectively. Consistent with our findings, similar increases in bole moisture during the growing season using in situ measurements has been reported in Douglas-fir (Beedlow et al 2007b) and other species (Hao et al 2013, Irvine et al. 1998, Wullschleger et al. 1996).

It is somewhat counter intuitive that bole RWC was minimal during winter when ASW was high, and increased as ASW was being drawn down in the spring and early summer. However, our results suggest the pattern is consistent with mechanisms associated with hydraulic dynamics in the sapwood and bark as influenced by growth and ASW.

Water in the sapwood, representing a substantial portion of a tree’s stored water, is released primarily by cavitation although a small fraction of this water is found in living tissues and is subject to elastic shrink-swell with changes in xylem water potential (Ćermák et al. 2007). Movement between the extensible tissues of the living bark and xylem follows a water potential gradient wherein movement results in shrinkage and swelling of the stem driven by changing water potential in the xylem. In conjunction with studies using sap flow measurements, the ability of living tissues to store water, which can contribute to the transpiration in times of water stress, has been established (e.g. Ćermák et al. 2007, Steppe et al. 2006, Zweifel et al. 2001, Zweifel et al., 2007). To our knowledge, no studies have included directly measured changes in bole water content in conjunction with stem diameter changes and sap flow to give a more complete picture of tree water dynamics. In this study we did not have sap flow data, nor did we have automated dendrometer data necessary to address this issue with better resolution. However, water loss from bark was assumed to occur during periods of stem shrinkage (negative rBAI) in late summer and winter.

Over the course of this study, substantial increases in the amount of cortex (OB′) in contact with the ThetaProbe rods were seen in some of the trees (Figure 3). The thickness of OB′ did affect the amount of water measured as seen in the increasing trend of RWC over the study period (Figures 1, 2 and 3). We can infer from Figures 1 and 2 that the bark does not affect the seasonal pattern of RWC because the amplitude of the annual RWC pattern does not differ markedly from the beginning of the study to the end. At the beginning OB′ was minimal (1–2 mm) because we removed it with a drill (see Methods). By the end of the study the OB′ increased in thickness by as much as 36mm covering up to 25% of the probe rods. Over the study period the thickness of the phloem remained relatively constant. Notwithstanding, the amplitude of the RWC fluctuations between winter and summer changed little if at all. Further the amplitude of the seasonal fluctuations was similar between trees with thick OB′ and those where it was much thinner.

Bole RWC is a function of transpiration and water uptake, which are driven by air and soil temperature, VPD, PAR and ASW (Ćermák et al. 2007, Lassoie 1982). In this study RWC was strongly correlated with those environmental factors over most of the annual cycle. With the onset of growth in the spring, RWC developed a strong, positive correlation with air and soil temperature, VPD, and PAR. ASW began to decline typically in May, but in some years as early as late March or as late as June, depending on snow cover. During this period air temperature was below the optimum for Douglas-fir and ASW was above 50% (Beedlow et al. 2013)—conditions where a positive response of RWC to factors favoring transpiration and sap flow are expected. In July and August, air temperature often exceeded the optimum of ~20 °C for radial expansion at this site, and the importance of VPD diminished suggesting longer periods of stomatal closure (Beedlow et al. 2013). These observations are consistent with the view that stomatal control during periods of drying soils and reduced root uptake of water, maintains xylem water potential above a critical threshold, which reduces embolism formation and minimizes irreversible damage to the tree’s hydrologic continuity (Bond and Kavanagh 1999, Cruiziat et al. 2002, Domec et al. 2004, Irvine et al. 1998, Kavanagh et al. 1999, Martínez-Vilalta et al. 2004, Utsumi et al. 2003). Further evidence for such stomatal control of RWC is the finding that sapwood RWC in Douglas-fir is higher at more xeric sites where more stomatal closure would be expected during summer drought (Beedlow et al. 2007a).

Measurements of _rBAI provide information on the timing of annual growth and changes in phenology. Based on Beedlow et al. (2007b), the annual pattern of rapidly increasing RWC from spring to summer is initiated just prior to the phenological phase of bud swell in the spring, and continues increasing through the sharp rise in _rBAI during bud break and earlywood formation. Earlywood is responsible for conducting most of the water through the bole (Domec and Gartner 2002b). Whenever there is growth, the new cells will be full of water when they are formed. While contributing to the increase in bole water content, new cell formation does not really answer the question of why RWC is highest when the soil is the driest. Although new cells could contribute to increasing bole water during active growth, it seems unlikely that the narrow band of new cells, relative to all the tissues contacted by the probe rods, would have a substantial effect on RWC in and of itself.

Peak _rBAI corresponds to the completion of shoot elongation and the time when the new needles become fully functional (Beedlow et al. 2007b). RWC reaches its highest levels after peak _rBAI before the start of latewood formation in August. Based on patterns of _rBAI, RWC begins decreasing after the shoots are expanded and during transition and latewood formation (Beedlow et al. 2007b). The onset of water stress limits shoot growth and is signaled by the formation of transition and latewood (Domec and Gartner 2002b, Lassoie 1982). Cambial activity and _rBAI slow as ASW decreases to minimum levels in summer (Beedlow et al. 2007b).

The decline in RWC during late summer and fall coincident with soil drying seen in this study suggests that bole water was being used in transpiration without complete recharge during the night. This progressive decrease is consistent with the hypothesis that during prolonged dry periods, transpiration exceeds water uptake on a diel basis resulting in a net loss of bole water, which over successive days or weeks is manifested in decreasing bole water storage (Ćermák et al. 2007, Phillips et al. 2003, Waring and Running 1978). The mechanism for this decrease is thought to be embolism formation in sapwood and the loss of water from elastic tissues (Ćermák et al. 2007, Domec and Gartner 2001, 2002a, 2002b, Irving and Grace 1997). Embolisms result in decreasing bole water content during drought periods in Betula papyrifera as measured by frequency domain moisture sensors, which operate on the same principle as the ThetaProbe sensors used in this study (Hao et al. 2013).

Though ASW is not the primary limiting factor to Douglas-fir growth at this site (Beedlow et al. 2013), a progressive decrease in bole water content was observed each year throughout the ten years of this study as ASW fell to 40–50 % in the upper 0.6 m of soil. Granier et al. (1999) suggest that water stress begins when 40% of plant available water remains in the soil. Previous studies have shown that Douglas-fir, a shallow rooted species, increases the use of deeper groundwater during mid-summer when precipitation inputs are low (Andrews et al. 2012, Meinzer et al. 2007, Warren et al. 2005), but that sap flux begins to fall rapidly once approximately 50% of available water in upper 0.6 m of soil is used (Warren et al., 2005). The appearance of diel RWC patterns in both the boles and canopies as RWC approached peak levels in summer and ASW fell below ~50% suggests that bole water was used during the day and recharged during the night. The progressive decrease in RWC during the summer drought indicates that nighttime recharge was insufficient to meet daytime transpiration needs.

Fall and winter can be critical for carbohydrate accumulation in PNW conifers because air temperature is often high enough to allow photosynthesis. As much as 50% of the annual carbon assimilation in Douglas-fir can occur between the months of October and May (Emmingham and Waring 1977). Though RWC decreased throughout fall and remained low in winter, it was positively correlated with air and soil temperature, VPD, and PAR, although variability was high as would be expected in intermittently photosynthesizing trees. RWC reached minimum values in late winter. McCulloh et al. (2011) report that loss of hydraulic conductivity in Douglas-fir due to embolisms is higher in winter than during the summer drought. In that study, they attribute the increase in winter embolisms to multiple freeze-thaw cycles; cold soil temperature (<9 °C) slowed or limited re-filling of embolized tracheids until spring. They conclude that for Douglas-fir, winter months are more stressful in terms of hydraulic functioning than the summer months. Cold soil inhibits water uptake, and if air temperature is high enough to allow transpiration, water stress can result (Aroca et al. 2011, García-Tejera et al. 2016, Kozlowski 1987). Warm periods are common during winter in Douglas-fir forests. In support the hypothesis that relatively low RWC during winter observed in this study is the result of water stress, temporary stem shrinkage has been observed during warm, sunny periods in winter indicating loss of water from bark tissues (Beedlow et al. 2013). Further evidence of soil temperature affecting RWC can be seen in 2010 (Figure 5). In that year soil temperature was relatively warm (~10 °C) and RWC stayed relatively high, a similar, but less pronounced event occurred in 2014. In short, low RWC in winter may result from a combination of freeze-thaw-induced embolisms in the sapwood and loss of water from bark tissues, indicated by stem shrinkage. Rehydration in both sapwood and bark is limited by low soil temperature.

Douglas-fir are thought to refill embolized sapwood both diurnally and seasonally (Ćermák et al. 2007, Hao et al. 2013). Seasonally, Hao et al. (2013) using frequency domain moisture sensors, found that increasing bole water content coincided with increasing root pressure and embolism repair during the spring. At our site the soil begins to warm during May, while ASW was at the growing season maximum. Prior to that time the ground was typically covered with snow and soil temperature in the root zone was below ~6 °C, which restricts root uptake of soil water (Aroca et al. 2011, García-Tejera et al. 2016, Kozlowski 1987, Lopushinsky and Kaufmann 1984). We found rapid RWC increases starting in May and hypothesize that this marks the period of embolism repair and bole water recharge that continues through the period of active growth and ends in mid-summer as latewood begins to form and ASW falls below approximately 50% of maximum.

In summary, we hypothesize that the observed seasonal pattern of peak RWC in summer as soil water is being drawn down, and minimal RWC during winter when soil water is maximal is the result of the dynamic interaction between water loss through transpiration and water uptake by the roots mediated by soil temperature. RWC declines when roots cannot absorb water fast enough to compensate for transpirational loss. Under conditions of low transpiration and high ASW, such as when soil is warming in the spring, RWC shows no diel pattern and increases day-to-day as sapwood and bark tissues are re-hydrated (Figure 4). The observed increasing RWC during late spring when soil water is decreasing is counter intuitive. We hypothesize that this is because that during that time when bole water is increasing, soil water is not yet limiting. Only as soil water continues to decrease below 50% as the summer drought wears on does it become limiting enough to affect bole water by slowing recharge to the point where there is a net loss from day to day. During periods of high transpiration in early summer when soil water is still adequate, RWC is drawn down during the daytime, but it is fully recharged during the nighttime with little to no net day-to-day change. While we do not have high resolution dendrometer data necessary to resolve diel shrink-swell patterns, we can clearly see RWC patterns of water loss during the daytime and rehydration during the. In mid- to late summer, progressive soil water decline below ~50% results in a slowing of water uptake by roots, incomplete bole water recharge during the night and decreasing RWC from one day to the next. This is reflected in stem shrinkage (negative rBAI during very warm, dry periods). If the water stress is great enough, embolisms can occur further lowering bole water content, although stomatal control in Douglas-fir has been shown to control embolism formation during drought to some extent. This pattern of decreasing RWC continues until the fall rains arrive, but decreasing soil temperature slows root uptake of water and, consequently, rehydration. During winter, the soil water is recharged through rain and snow melt, but cold soil inhibits root uptake of water. Freeze-thaw-induced embolisms can further reduce RWC. During sunny periods in winter, tree canopies often become warm enough to photosynthesize. Transpiration with limited water uptake by cold roots also contributes to low winter-time RWC. Bole water content does not increase significantly until the soil warms in the spring. With warming of the soil, water uptake increases to match transpiration, re-filling embolized tracheids and rehydrating bark tissues resulting in a marked increase in RWC. Increasing RWC continues with growth until the soil water again becomes limiting in mid-summer.

5. Conclusions

Seasonal patterns of bole water content in large, old Douglas fir trees (400+ years) are similar to younger trees (~150 and ~115 years). Changes in bole RWC from winter to summer are small relative to changes in soil moisture. Maximum RWC occurred in mid-summer when soil moisture was still above 40–50% of maximum levels. The lowest bole RWC occurred during winter when ASW peaked. The seasonal patterns of RWC are consistent with phenological patterns of shoot and cambial development. RWC was strongly correlated with environmental factors that govern transpiration and water uptake over most of the annual cycle. The progressive decrease in RWC from mid-summer through fall in this study supports the hypothesis that bole water is increasingly used in transpiration during prolonged dry periods, or during periods of active transpiration when low soil temperature inhibits water uptake by roots. We hypothesize that the increase in RWC during spring and early summer coincident with active growth, marks the period of embolism repair and bole water recharge. We conclude that bole water storage is an integral part of tree water dynamics enabling trees to extend carbon assimilation into drought periods and over winter, an adaptation that could benefit the survival of large PNW trees should climate change prolong and intensify the annual summer drought while increasing winter temperature.