The pruning trial was established in a research plantation near Nana Glen in northeastern New South Wales (30°1′S, 153°8′E), where both species were planted in adjacent stands. This well-drained site contains gently sloping deep (1–1.5 m) brown and yellow earths soils [20] derived from late carboniferous siltstone, mudstone and conglomerate [21]. The site is approximately 165 m above sea level and receives a moderately high annual precipitation of 1437 mm (measured 1920–2004), distributed with a distinct winter minimum and summer/autumn maximum.

Original site vegetation consisted of a tall open mixed hardwood forest including E. pilularis, E. intermedia R. Baker and E. microcorys F. Muell. The site was cleared early last century and subsequently converted to pasture for grazing. To prepare the site for planting, the soil was ripped in a north-south direction in September 2000 to a depth of 0.7 m and mounded to obtain 4 m wide row spacings. A second cultivation was completed one month later to decrease soil tilth for planting. Herbicides (glyphosate 4 L ha⁻¹, simazine 2.5 kg ha⁻¹ and metolachlor 1.5 L ha⁻¹) were applied to mounded soil one month prior to planting in December 2000. Eucalyptus pilularis (Whian Whian State Forest seedlot) and E. cloeziana (Pomona State Park and Mebbin State Forest plantation seedlot) were planted at a stocking of 1250 trees ha⁻¹. Each seedling was fertilised with 9 g elemental nitrogen and 10 g elemental phosphorus in the form of diammonium phosphate at the time of planting. Post-plant weed control involved applications of haloxyfop (0.5 L ha⁻¹) and clopyralid (0.8 L ha⁻¹) to planted mounds one and four months after planting.

Six 30 × 30 m blocks of each species were located on a gentle east-facing slope (<4°). Within each block, two trees were selected with a similar diameter at breast height (1.3 m, DBH). These 12 trees per species were of the co-dominant or dominant crown class within the stand [22], straight and single stemmed, free of visible health defects and surrounded by four immediate neighbours in each direction. On the 6 October 2004, one randomly selected tree in each block was pruned to remove the lower 50% of the length of the live-crown. Live-crown length was visually defined as the distance between the tree height and the stem insertion height of the lowest live branch contained within a geometrically regular crown envelope [23]. In practice, pruning regimes for these species are being developed but it is likely that all pruning up to a height of about 5–6 m would be completed by about age five or six years.

The twelve selected trees per species were monitored for 87 days with sixteen sap flow sensor units (Model SF100 Greenspan Technology, Warwick, Queensland, Australia). Units were deployed for 12 days before pruning (24 September to 5 October 2004) and 75 days after pruning (7 October to 20 December 2004) using a roaming sensor technique [24,25,26]. Trees were not monitored on the day (6 October 2004) on which pruning occurred. A total of four trees (from the 12 selected trees per species), one unpruned and one pruned per species, were randomly selected as reference trees. Two sensor units (four probesets) were used to take measurements on the northern, southern, eastern and western side of the stem of each reference tree. Thermistor pairs were placed 5 and 10 mm under the cambium in northern and western aspects and 10 and 15 mm under the cambium in the southern and eastern aspects. The remaining eight roaming sensor units were moved from tree to tree among the remaining 20 trees every 4 to 13 days. On each roaming sensor tree, one sensor unit (two probesets) was used, with Probeset 1 placed on the northern side with thermistor pairs 5 and 10 mm under the cambium and Probeset 2 on the eastern side of the tree with thermistor pairs 10 and 15 mm under the cambium. SF300 Model sensors (Greenspan Technology, Warwick, Queensland, Australia) were used to replace problem sensors. This involved placing one set of sensors to obtain sap velocity measurements 10 mm under the cambium. Since the replacement sensors could only provide data from an insertion depth of 10 mm under the cambium, only data collected from this depth were used in the final analysis.

Each of the 20 roaming sensor trees received a minimum of 4 monitoring days prior to pruning and 12 monitoring days post pruning. The four reference trees however, were measured continuously throughout the entire 87 day monitoring period. This technique of sensor deployment allows measurements of water use obtained from the roaming sensors to be regressed against those obtained with the reference sensors [24,25,26] before and after pruning. The regression equations were used to predict water use in roaming sensor trees outside the period of measurement in those trees thus enabling replication across the entire measurement period [24,25,26].

All probesets were inserted 0.12 m above ground and no live branch, pruned branch wound or stem deformity occurred within 0.12 m above or below the insertion point. To avoid the onset of wound healing during the measurement period, toothpicks greased with vaseline were placed in thermistor holes when trees were not being monitored. To avoid excessive wounding effects, the probes on the reference trees were reinstalled at intervals of 28 days at a different position at the same height. All probesets were wrapped in aluminium foil to reflect direct radiation. Heat pulses were applied for 1.8 s at 20 min intervals. Heat pulse times greater than 180 s were regarded as indicating a zero rate of sap flow. This insensitivity to slow flow rates (>180 s) is a disadvantage of this equipment, however the total transpiration and hence the higher transpiration rates were of more interest in this study. Values of heat pulse velocity were corrected for wounding by assuming a wound diameter of 2.2 mm [26].

Radial variation in sapwood conductivity was accounted for by completing sap velocity profiles for each tree at the completion of the experiment [27]. Radial profiles were obtained on the southern side of each tree by moving a sensor in 5 mm increments 1 min before heat pulse (10 min heat pulse interval) on warm clear days. This profile was used in conjunction with point measurements obtained from the northern and eastern side of the tree to determine a correction coefficient (a mean of point to moving sap velocity ratios, weighed by area with depth in the sapwood) for routine point measurements [27]. This method follows recommendations from Zang et al. [27] and applied by Hunt and Beadle [25] that sampling on multiple sides was not necessary to obtain accurate estimates of sap velocity in young Eucalyptus trees. Correction coefficients for the northern and eastern stationary sensors were applied to routine measurements of sap velocity made 10 mm under the cambium.

At the completion of the experiment two cores were removed from each tree on the northern and eastern aspects using a 5 mm increment corer to calculate sapwood area. Bark thickness was measured and the radius of sapwood on each aspect was determined by staining the core at the heart wood boundary with 5% solution of dimethyl orange indicator. Unstained core sections from each aspect were used for gravimetric determination of the volume fractions of wood (V_w) and water (V_h) of each tree.

To assess changes in tree growth over the monitoring period, measurements of tree height, diameter at breast height (1.3 m, DBH) and stem diameter at insertion height (0.12 m) were completed at the beginning and end of the measurement period on all trees. Height to the live crown base and the four crown radii within and between rows were measured at the beginning of the measurement period to characterise crown dimensions. Height to the live crown base was subtracted from total height to calculate live-crown length. The mean crown radii within and between rows were used to calculate projected crown area using the area of an ellipse.

Measurements of rainfall were made using a 0.2 mm tipping-bucket rain gauge with a bucket located 200 m from the experimental site. A Starlogger 6003B Portable Data Logger (Unidata Pty Ltd., Perth, Western Australia, Australia) was attached and programmed to record total rainfall at 15 min intervals. Maximum and minimum temperatures, mean daily EPan and VPD_max were derived from interpolated climate surfaces for the 88 day experimental period [28]. Average daily maximum and minimum temperatures were between 23.6 and 13.5 °C (1957–2004), with low minimum temperatures confined to winter months. Maximum vapour pressure deficit was derived from the saturated vapour pressure function of Buck [29] using modelled maximum daily temperature and modelled vapour pressure temperature [28]. Average daily pan evaporation (EPan) varied from about 2.8 mm in winter months to 5.9 mm in summer months (1957–2004).

Mean daily sap velocity (mm·s⁻¹) was calculated for each 24 h period. Sap velocity multiplied by sapwood area was used to determine mean daily transpiration (L day⁻¹) in individual trees. Mean daily sap velocity and transpiration for roaming sensor trees were regressed against values obtained from their respective reference trees to estimate mean daily transpiration and mean daily sap velocity of sample trees for the pre- and post-pruning measurement periods [24,25,26]. Significant regressions between mean daily transpiration and mean daily sap velocity were found between all reference and roaming sample trees of both species before and after pruning. Coefficients of determination (r² values) ranged from 0.56 to 0.98 before pruning and from 0.73 to 0.96 after pruning.

One-way analysis of variance (ANOVA) was used to assess treatment differences within a species in projected crown area, live crown length, mean daily transpiration and mean daily sap velocity before pruning. Treatment differences in DBH, stem area and tree height before and after pruning and sapwood area and width at probe insertion height and the volume fractions of wood and water at the end of the experiment were also assessed using one-way ANOVA.

Examining the data from all trees over the course of the experiment, it was evident that little difference between treatment trees was observed during days of rain and low VPD_max. To assess changes in mean daily transpiration over the course of the experiment in more detail, it was necessary to consider conditions where the differences in transpiration between treatment trees were highest. Six periods of fine weather (one six-day period pre-pruning and five eight-day periods post-pruning) were selected and mean daily transpiration of unpruned and pruned trees analyzed for both species.

Repeated measures ANOVA was used to assess the effects of pruning treatment on mean daily transpiration during the six periods of fine weather (one pre-pruning period and five post-pruning periods). Treatment, time and the interaction of treatment and time were included in the model. All analyses were performed in Genstat (VSN International 2004, Hemel Hempstead, Herts, UK). Post-hoc tests were made using least significant difference.

Before pruning, the DBH ranges were 10.2 to 11.9 cm for E. pilularis trees and 10.9 to 13.8 cm for E. cloeziana trees. The maximum difference in DBH between trees within any single block was 1.3 cm for E. pilularis and 1.4 cm for E. cloeziana. At the start of the experiment, DBH, tree basal area, cross-sectional area at the height of probe insertion, tree height, projected crown area and live-crown length, did not differ significantly between treatments (Table 1). There were still no pruning effects on tree size or growth rates at the completion of the experiment. Pruning did not significantly affect the volumetric fractions of wood or water in either species (Table 1). The sapwood area of one E. cloeziana tree suggested that it was an outlier so the two trees from that block were excluded from all analyses.

To explore the effects of initial tree size on transpiration in both species, mean daily measurements of transpiration were determined for all trees for the 12-day pre-pruning measurement period. Mean daily transpiration was 12 L day⁻¹ for both pruned and unpruned E. pilularis trees during the 12-day pre-pruning measurement period. Higher rates of mean daily transpiration were found in both treatments of E. cloeziana than E. pilularis, with 16 L day⁻¹ measured in unpruned trees and 18 L day⁻¹ measured in pruned trees during the same 12-day pre-pruning period. No significant difference in mean daily transpiration was found between pruned and unpruned trees of either species before pruning took place.

Sap velocity was higher between 5 and 15 mm under the cambium in all sample trees. Comparisons of sap velocity between the northern and eastern aspect of stems showed that values were generally highest for the eastern aspect but mean correction coefficients calculated from radial profiles (Figure 1) for the northern and eastern aspects at a depth of 10 mm under the cambium were not significantly affected by pruning (Table 2). Nor were there significant differences between correction coefficients calculated for the northern and eastern aspects for either species (Table 2).

Mean daily transpiration in the individual unpruned and pruned E. pilularis reference trees were 9 L day⁻¹ and 11 L day⁻¹, respectively, over the 87 day measurement period (Figure 2a). For this same period, mean daily transpiration in E. cloeziana was 15 L day⁻¹ in the unpruned reference tree and 16 L day⁻¹ in the pruned reference tree (Figure 2b). Values of mean daily transpiration between treatments of both species were reduced during days of wet weather or reduced temperatures (days 24 to 28, 59 to 61 and 75 to 79) (Figure 2). During days of fine weather and high temperatures, the difference between treatments of both species was magnified (days 14 to 21, 49 to 57 and 80 to 87). Following pruning, the pruned reference trees of both species showed reductions in mean daily transpiration (Figure 2). Following the 198 mm of precipitation between days 24 and 28, the difference in transpiration between pruned and unpruned trees appeared to be reduced in both species. Approximately 37 days after pruning (day 50), the two pruned reference trees had higher rates of transpiration than their respective unpruned counterpart in both species, particularly during periods of fine weather. During the initial 37 day (days 14 to 50) post-pruning monitoring period, mean daily transpiration varied between 2 and 14 L day⁻¹ in the unpruned and 3 and 15 L day⁻¹ in the pruned E. pilularis reference trees. Similar levels of variation were still present in the final 38 days (days 51 to 88), with transpiration values between 4 and 11 L day⁻¹ in the unpruned reference tree and between 5 and 15 L day⁻¹ in pruned E. pilularis reference tree. Mean daily transpiration was more variable between the unpruned and pruned E. cloeziana reference trees (1 and 20 L day⁻¹) than between unpruned and pruned E. pilularis reference trees (3 and 19 L day⁻¹) during the initial 37 day post-pruning period. Similarly during the final 38 day post-pruning period mean daily transpiration was more variable in E. cloeziana reference trees (11 and 29 L day⁻¹) than E. pilualris (7 and 22 L day⁻¹) reference trees.

Significant differences in mean daily transpiration were found between treatments across the six periods of fine weather in both species (Figure 3, Table 3). Prior to pruning, mean daily transpiration was not significantly different between pruned and unpruned trees in either species (Figure 3). However, during the two eight-day periods immediately following pruning, mean transpiration was significantly lower in pruned than unpruned trees of both species. Pruning reduced mean daily water use by 39% in E. pilularis and 59% in E. cloeziana compared to the unpruned trees during the first eight-days immediately following pruning (days 14–21), and 30% in E. pilularis and 44% in E. cloeziana in the second eight-day period after pruning (days 33–40). Thirty six to forty three days (days 49–56) after pruning and during subsequent fine weather periods, mean daily transpiration was not significantly different between pruned and unpruned trees.

Total rainfall of 27 mm before pruning and 339 mm after pruning was recorded during the experiment (Figure 2d). Daily minimum and maximum temperatures ranged between 9 and 40 °C. VPD_max averaged 1.6 kPa for the experimental period but did reach 5.6 kPa on one day when maximum temperature was 40 °C (Figure 2c). Daily EPan on this day was correspondingly high at 10.2 mm but averaged 5.2 mm over the experimental period.