3.4 Effect of girdling on radial dynamics in bark and xylem
Phloem girdling markedly impacted radial fluctuations in xylem (dD x) and bark tissues (dD b + dD Gempir) below and above the girdle (Fig. 4). Prior to girdling (6-10 May), average dD x was 22.6 µm d-1 (± 5.0) at the base and 22.2 d-1 µm d-1 (± 7.9) at the top of trees. dD b + dD Gempir was nearly four times larger compared to dD x, measuring 76.8 µm d-1 (± 29.8, ± 1 SD) at the base and 71.8 d-1 µm d-1 (± 13.5) at the top (Fig. 4a, b). Importantly, and independent of girdling, the dynamic changes of dD x and dD b + dD Gempir were asynchronous, whereby dD x exhibited the largest expansion around 14:00 h when dD b + dD Gempir reached its minimum.
Following girdling (20-24 May), dD x at the top of stems (above the girdle) remained unchanged (21.9 µm d-1 ± 7.9), but significantly declined below the girdle (13.2 µm d-1 ± 9.7; p <0.001) (Fig. 4c, d). Girdling slightly increased dD b + dD Gempir above the girdle (82.3 µm d-1 ± 26.0) while it was reduced by more than half below the girdle (33.5 µm d-1 ± 10.7; p <0.001). dD Gempir was similar at both measurement positions (base: 49.5 µm d-1 ± 29.7; top: 51.6 d-1 µm d-1 ± 17.6) before trees were girdled. Post-girdling, dD Gempir ceased at the base (0.2 µm d-1 ± 8.5) yet slightly increased at the top (80.1 µm d-1 ± 28.9).
Interpolated cumulative D Gempir(empirical data based on daily minima) as well as modelledD G+ at the base of trees showed a continuous increase of plastic growth prior to girdling with no distinct diurnal pattern (Fig. 4e). During daytime (7:00-19:00 h) meanD Gempir was 52% (±4.5) andD G+ was 50.3% (±2.7) of total plastic growth realised during a 24-hour cycle (i.e., 7:00-7:00 h). As indicated earlier, plastic growth below the girdle ceased entirely, thus no diel patterns were identifiable. At the top of trees, plastic growth was also continuous before the gridle was applied, yet at this positionD Gempir andD G+ showed a distinct diel cycle with high plastic growth during the day (Fig. 4f). Between 7:00 h and 19:00 h,D Gempir accounted for 77.2% (±6.8) and D G+ for 77.6% (±1.0) of total plastic growth within the 24-hour cycle. After trees were girdled, this proportion declined to 62.5% forD Gempir and 71.9% forD G+, retaining a clear diurnal rhythm of high plastic growth during the day and lower plastic growth during the night.
Our improved model reproduced the magnitude of plastic growth, stripped of hydraulic and osmotic signals (D G+), well for the base and top measurement positions, and also before and after girdling when compared toD Gempir (Fig. 5). During the 5-day interval before girdling, average cumulativeD Gempir was 258.2 µm at the base and 256.2 µm at the top. Model predictions for the same time interval were similar at the base (287.2 µm), but slightly higher at the top (346.3 µm). After girdling of trees, the improved model matched empirical measurements well, reflecting the collapse of growth below the girdle and an increase of D G+ at the top (Fig. 5).
Separating \(\frac{{dD}_{b}^{\text{TOT}}}{d_{t}}\) into components 1, 2 and 3 using empirical measurements from dendrometers, sap flow, leaf water potential and environmental information allowed us to separate simultaneously occurring hydraulic (Fig. 6) and osmotic processes (Fig. 7) in live bark, the cambial growth zone and xylem during different parts of the diel cycle. First, model outputs documented systematic fluctuations in dD b and dD x in response to diel changes in hydrostatic pressure in the xylem (Fig. 6a, b). With the onset of transpiration in the morning, pressure potential in the xylem (dD xΨP,x) declined, leading to a small (approx. 3 µm) but systematic contraction of sapwood. The increasing negative pressure in xylem vessels was transduced via rays into the live bark, where after a short lag time capacitive water stores began to empty water into the xylem, leading to a large contraction of live bark tissue (dD bΨP,x). Following peak transpiration during midday (see Fig. S3), negative pressure in xylem eased, yet capacitive water continued to be drawn from live bark tissue leading to further contraction of live bark, leaving a total daytime contraction of up to 60 µm. With the onset of darkness, transpiration rates were low, allowing dD xΨP,x to fully recover and remain largely unchanged during the night. At the same time, capacitive stores were recharged leading to an increase of dD bΨP,x. Prior to girdling, the magnitude of these hydrostatic processes did not vary between the measurement positions, producing a sine-type oscillation of dD bΨP,x.
After trees were girdled, the amplitude of dD xΨP,x and dD bΨP,x markedly declined (Fig. 6c, d). dD xΨP,x was reduced to less than 1µm while its distinct transpiration-driven phases of decline, recovery, and steady-state, remained below and above the girdle. The principal sine-wave pattern of viscoelastic changes in live bark also remained, albeit reduced from 60 to around 20 µm after girdling and at both measurement positions. However, model outputs indicated that girdling caused dD bΨP,x below the girdle to deviate from the oscillating pattern around zero to one that was increasingly positive (Fig. 6c), indicating a ‘swelling’ of live bark tissue. This was not observed for dD bΨP,x above the girdle.
Prior to girdling, our model showed a simultaneous decline in dD bΨP,x (Fig. 6) andD xΨΠ,x (Fig. 7a, b), with similar dimensional change at the base (up to 60 µm) but not the top of trees (15-20 µm). Similar to the oscillating sine-pattern of dD bΨP,x,D xΨΠ,x at the base of trees recovered overnight, peaking during early morning. However,D xΨΠ,x at the top of trees showed no recovery during the night, but a sharp increase during the early morning before and also after trees were girdled. This pattern was very similar to dendrometer measurements of D xshown in Fig. 4b and 4d. Finally, expansion of live bark as result of changes in osmotic pressure within this tissue (D bΨΠ,b) peaked at the end of every day before girdling at the base and top of trees (Fig. 7c, d). After girdling, these peaks disappeared below the girdle andD bΨΠ,b steadily declined, showing that expansion of D b was increasingly less driven by osmotic pressure gradients. The opposite trend, a constant increase in D bΨΠ,b was observed above the girdle. At both positions, the typical late afternoon peak observed prior to girdling had disappeared post-girdling.
Our mixed effect model, that accounted for more than one distinct source of variability, was able to explain a large proportion of uncertainty (conditional R2 was 0.47 before and 0.40 after girdling; marginal R2 was 0.51 before girdling and 0.47 after girdling; Table 3). The observed dimensional changes inD x and D b resulting from viscoelastic changes in live bark and phloem, as well as changes in patterns of D G+ are in agreement with effects associated with reduced transpiration (Fig. S3) and changes in the concentration of NSC that affect physiological and osmotic dynamics (Fig. 1), below and above the girdle.