1

Drought stress physiology of evergreen sclerophyllous broad-leaved trees assessed with the non-

invasive magnetic turgor pressure probe

Bader Martin K-F1,6*, Ehrenberger W2,4, Bitter R2,4, Stevens J3, Miller B3, Chopard J1, Rüger S4, Hardy

G5, Poot P1, Dixon KW3, Zimmermann U4 and Veneklaas EJ1

1School of Plant Biology and Centre of Excellence for Climate Change, Woodland & Forest Health,

University of Western Australia, 35 Stirling Highway, Crawley WA 6009, Australia 2University of Würzburg, Department of Biotechnology, Am Hubland, 97074 Würzburg, Germany 3Kings Park and Botanic Garden, Fraser Avenue, West Perth 6005 WA, Australia; School of Plant

Biology, University of Western Australia, Crawley 6009 WA, Australia 4ZIM Plant Technology, Neuendorfstr. 19, 16761 Hennigsdorf, Germany 5School of Biological Sciences and Biotechnology and Centre of Excellence for Climate Change,

Woodland and Forest Health, Murdoch University, South Street, Murdoch WA 6150, Australia 6New Zealand Forest Research Institute (SCION), Te Papa Tipu Innovation Park, 49 Sala Street, Private

Bag 3020, Rotorua 3010, New Zealand

*Author to whom correspondence should be sent:

Martin.Bader@scionresearch.com | New Zealand Forest Research Institute (SCION), Te Papa Tipu

Innovation Park, 49 Sala Street, Private Bag 3020, Rotorua 3010, New Zealand

Author contributions: MKFB and UZ conceived and designed the experiments. MKFB, JC, JS and BM

performed the experiments. MKFB and WE analysed the data. MKFB and UZ wrote the manuscript;

EJV and GH won the overarching grant; other authors provided editorial advice.

Abstract 1 2 Southwest Australian Banksia woodlands are highly diverse plant communities that are threatened by 3 drought- or temperature-induced mortality due to region’s changing climate. We examined, water 4 relations in Banksia menziesii R. Br. using a novel magnetic leaf patch clamp pressure probe (ZIM-5 probe) that allows continuous, real-time monitoring of leaf water status. Multiple ZIM-probes across the 6 crown were complemented by traditional ecophysiological measurements. During summer, early 7 stomatal down-regulation of transpiration prevented midday balancing pressures from exceeding 2.5 8 MPa. Diurnal patterns of ZIM-probe and pressure chamber readings agreed reasonably well, however, 9 ZIM-probes revealed short-term dynamics and leaf internal turgor gradients, which are impossible to 10 capture using a pressure chamber. Drought stress manifested as increasing daily amplitudes and 11 occasionally as rising baseline at night. These symptoms occurred more often locally than across the 12 entire crown. Microclimate effects on leaf water status were strongest in the East and North. Extreme 13 spring temperatures preceded the sudden death of B. menziesii trees, suggesting a temperature- or 14 humidity-related tipping point causing rapid hydraulic failure as evidenced by collapsing ZIM-probe 15 readings from an affected tree. In a warmer and drier future, local mortality events in B. menziesii 16 populations may occur more frequently with strong impacts on community structure and ecosystem 17 functioning. 18 19 Key words: Leaf turgor, water relations, Banksia, climate change, Mediterranean ecosystems 20 Running headline: Non-invasive real-time drought stress monitoring 21 22 Introduction 23 24 The Mediterranean biome is characterised by an extraordinarily high level of plant diversity and 25 endemism, which has been attributed to adaptation to the typical climate with dry hot summers and cool 26 wet winters, frequent fires and nutrient-impoverished soils (Breckle 2002). Plants have evolved a 27 remarkable suite of phenological, physiological and morphological adaptations to cope with these harsh 28 conditions such as sclerophyllous leaves, lignotubers and epicormic buds for regeneration after fire or 29 cluster roots to enhance nutrient acquisition from the phosphorus-deficient soils (Lambers et al. 2008). 30

2

However, despite these diverse adaptations, Mediterranean-type ecosystems are highly 31 vulnerable to climate change (Klausmeyer and Shaw 2009). A plant community that appears particularly 32 vulnerable to climatic drying is Banksia woodland (Sommer and Froend 2011) which is characteristic 33 of the Swan Coastal Plain – a ca. 30 km wide and 400 km long coastal strip that harbours the centre of 34 the Southwest Australian global biodiversity hotspot. Banksia menziesii R. Br. and Banksia attenuata 35 R. Br. (Proteaceae) co-dominate the overstorey of this highly diverse woodland (Keighery 2011) that 36 occurs on coarsely textured sandy soils with extremely low water-holding capacity which causes the soil 37 profile to dry out rapidly to considerable depths during the hot, dry summer period (Groom 2011). These 38 two dominant Banksia species are facultative phreatophytes (Dodd and Bell 1993; Canham et al. 2009) 39 and may access deep soil water resources and groundwater down to a maximum of ca. 10 m through 40 their deep-reaching dimorphic root system (Groom 2004). However, over the past few decades, 41 declining precipitation has resulted in receding groundwater tables and insufficient soil moisture 42 recharge, which has further exacerbated summer drought stress in this region (Timbal et al. 2006; 43 Hughes et al. 2012). In 2010, the southwest of Western Australia experienced its driest and one of the 44 hottest years on record (503.8 mm annual rainfall, Bureau of Meteorology 2011), causing extensive tree 45 mortality in forests and woodlands across the landscape. In Perth’s Kings Park urban bushland, it was 46 estimated that 60-70% of Banksia spp. populations succumbed to the severe heat and/or drought during 47 that summer (Miller and Easton, unpublished). 48

In this study, we investigated crown water relations of Banksia menziesii in response to spring 49 and summer drought in a system with documented significant change in rainfall, temperature and tree 50 cover (Crosti et al. 2007; Australian Bureau of Meteorology). For the first time, we present continuous 51 data on leaf water status across the crown of B. menziesii trees using the novel magnetic leaf patch clamp 52 pressure probe (trade name: ZIM-probe). We asked whether incipient drought stress could be reliably 53 detected in highly sclerophyllous trees and whether crown regions experiencing longer sun exposure 54 were more susceptible to drought. By combining real-time leaf water status measurements with 55 microclimate and sap flow recordings and a diurnal measurement campaign during mid-summer using 56 conventional ecophysiological techniques, we also aimed to contribute to the long-standing debate about 57 the complex spatial and temporal dynamics of water relations across multiple scales from within-leaf to 58 crown on a daily to seasonal basis (Huber and Schmidt 1936; Tobiessen et al. 1971; Connor et al. 1977; 59 Hinckley et al. 1978; Turner and Long 1980; Turner et al. 1984; Passioura 1991; Jones 1992; Canny 60 1993; Milburn 1996; Goldstein et al. 1998; Meinzer et al. 1999b; Clearwater and Clark 2003; 61 Zimmermann et al. 2004; Zweifel et al. 2005, 2007, 2009). 62 63 Materials and Methods 64 65 This study was conducted in the 267 ha native bushland reserve of Kings Park Botanic Gardens in Perth, 66 Western Australia (31.96°S, 115.83°E). The study area is part of the Spearwood Dune System with soils 67 of the Karrakatta soil association, consisting predominantly of nutrient-poor sands (McArthur 1991). 68 The Perth region is distinguished by a pronounced Mediterranean-type climate with hot dry summers 69 (December–March) and cool wet winters (see Fig. S1 available as supplementary material to this paper). 70 Mean annual precipitation amounts to 734 mm most of which falls between May and September. Mean 71 maximum and minimum air temperatures in summer (February) are 31.4 °C and 18.2 °C, respectively, 72 and in winter (July) 18.3 °C and 7.7 °C (Australian Bureau of Meteorology). 73

In early August 2011, a mature ca. 6-m-tall Banksia menziesii R. Br. (Proteaceae) tree – growing 74 at an elevated site where the groundwater level is well beyond maximum rooting depth – was initially 75 equipped with 6 magnetic leaf patch clamp pressure probes (trade name ZIM-probeTM, ZIM Plant 76 Technology GmbH, Hennigsdorf, Germany) for continuous monitoring of leaf water status. On 6 77 October, four additional leaf patch clamp pressure probes were installed. The tree was equipped with 2 78 probes in each cardinal direction on sunlit leaves in the upper and lower crown with the remaining two 79 probes located in the shaded centre of the crown and centrally on top of the crown. Because the 80 sclerophyllous leaves are clustered at the end of the branches, the establishment of a radial leaf profile 81 across the crown was not feasible. Canopy access was achieved by using a trailer mounted boom lift. 82 This study tree died in late October 2011, following several days of unusually high ambient temperature 83 (see below). On 22 December 2011, we equipped a new 5.5 m tall B. menziesii tree, initially with 8 84 pressure probes, one in the lower and one in the upper crown of each cardinal direction (on average ca. 85

3

1.5 m vertical distance from upper to lower crown). In February 2012, we increased the number of 86 probes from one to two in each measuring location. Moreover, we equipped one upper-crown leaf in 87 each cardinal direction with 3 leaf patch clamp pressure probes at the base, mid and tip of the blade to 88 test whether within-leaf turgor gradients exist. 89 90 The leaf patch clamp pressure probe 91 92 Principle 93 The principle of the magnetic leaf patch clamp pressure probe is described in detail elsewhere 94 (Zimmermann et al. 2008; Westhoff et al. 2009a; Fernández et al. 2011). Briefly, an intact leaf is 95 clamped between the two planar pads of the probe, each equipped with a toric magnet. One of the pads 96 contains a highly sensitive pressure sensor, the other one a threaded rod for varying the distance between 97 the two magnets. The adjustment of the distance between the two magnets allows the optimum setting 98 of the magnetic clamp pressure, Pclamp (that is kept constant during measurements). The small covered 99 leaf patch between the magnets is used as a sensing element for measuring turgor pressure changes in 100 the uncovered leaf surrounding. This implies that the patch is hydrated and in hydraulic (and osmotic) 101 contact with the entire leaf. 102

The turgor pressure, Pc in the leaf patch (given by the osmotic pressure difference between the 103 cellular and apoplastic compartments, is opposed to the magnetic pressure, Pclamp, applied to the leaf 104 patch. This means that the output patch pressure, Pp, sensed by the pressure sensor is low at high turgor 105 pressure and high at low turgor pressure, Pc, based on the physics of pressure transfer through the leaf 106 (Zimmermann et al. 2008): 107

(Eq. 1) 108 109

where Fa is a leaf-specific attenuation factor which takes into account the part of Pclamp that is 110 used for compression of air-filled spaces and structural elements such as cuticle and cell walls. Fa can 111 be assumed to be constant down to very low turgor pressures (ca. 50 kPa), even over long measuring 112 periods (Zimmermann et al. 2008; Westhoff et al. 2009a).For the derivation of Eq. 1 a linear dependence 113 of the bulk elastic modulus, ε of the leaf patch on Pc has been assumed (Zimmermann et al 2008): 114

𝜀 = 𝑎𝑃𝑐 + 𝑏 (Eq. 2) 115 116

The magnitude of the coefficients a and b depend on the elasticity of the patch tissue and of the rate 117 constant of turgor pressure changes (depending among other things on the hydraulic conductivity). This 118 can readily be shown if the theoretical considerations of Murphy and Ortega (1995) are taken into 119 account that describe more accurately the changes of the volumetric elastic modulus of the leaf patch on 120 rapid changes of turgor pressure than Eq. 2: 121 122 𝜀(𝑃𝑐) = 𝜀∞ − (𝜀∞ − 𝜀0)𝑒−𝑘𝑃𝑐 (Eq. 3) 123 124 where k is the rate constant, ε0 the volumetric elastic modulus at Pc = 0 and ε∞ the volumetric elastic 125 modulus at large turgor pressures. 126 127 Using Eq. 3 for the integration of the transfer function (see Zimmermann et al. 2008) shows that the 128 bracket in Eq. 1 is replaced by: 129 130

[𝜀0

𝜀0+ 𝜀∞ (𝑒𝑘𝑃𝑐−1)]

1

𝑘𝜀∞ (Eq. 4) 131

132 The exponential term in Eq. 4 can be approximated by a Taylor expansion which can be terminated after 133 1 + kPc for relatively small pressures (say c. 500 kPa or less). Replacing of the e-function in Eq. 4 by 134 the Taylor expansion leads to Eq. 1.It is obvious that b = ε0 and a = k ε∞. Patch pressure measurements, 135 therefore, give in principle access to the hydraulic conductivity and the volumetric elastic moduli at Pc 136 = 0 and at high Pc values if two of these parameters can directly be measured. This can only be performed 137

clampa

a

c

p PFbaP

bP

1

4

by the cell turgor pressure probe on an intact leaf what is not possible on Banksia leaves because of the 138 high rigidity. 139 140 Experimental setup 141 If not otherwise stated, all probes were attached to the mid section of leaves, avoiding the main vein. 142 The adaxial side of the leaves of B. menziesii is rather smooth, whereas the abaxial side is coarsely 143 structured by prominent veins. Homogenous contact between the leaf and the sensing area of the probe 144 is a mandatory prerequisite for accurate measurements (among other things, this was most likely a major 145 reason why previous attempts to measure turgor pressure by application of an external force failed; see 146 e.g. Heathcote et al. 1979; Turner and Sobrado 1983). To ensure optimum contact between sensor and 147 leaf patch, the sensor-holding part of the probe was placed on the adaxial leaf surface. Clamping the 148 sensor-holding part to the abaxial side was much more time-consuming because of the protruding leaf 149 venation, but gave identical results (data not shown; Zimmermann et al. 2008). Real-time probe readings 150 were sent wirelessly by telemetric transmitters (ZIM Plant Technology GmbH, Hennigsdorf, Germany) 151 every 5 min to a receiver base station which logged to a GPRS modem linked to an Internet server. 152

The ZIM-probes were virtually temperature-independent as shown by control experiments 153 where temperature changes over a range of 20 °C only resulted in a change of ca. 1–2 kPa. Compared 154 to the turgor pressure signals (see below) this noise level is sufficiently small to be neglected in the 155 analysis (Zimmermann et al. 2008; Westhoff et al. 2009a). 156

157 The magnetic probes measure relative changes of turgor pressure. To allow comparison of 158

diurnal Pp measurements the Pp data were offset corrected to Pp*. For the offset correction the minimum 159 Pp value of the second day after clamping of the respective probe was used. For further normalisation 160 this offset corrected patch pressure (Pp*) was divided by the maximum Pp reading during the respective 161 day (Pp*/ Pp, max* [%]). For clarity, Pp data were smoothed using a 9-point FFT Filter routine in Origin 162 8.5 (Microcal Software Inc., Northampton, MA, USA). 163 164 Microclimate measurements 165 166 Relative humidity and ambient temperature were recorded in parallel to the ZIM-probe measurements 167 using initially two sensors of each kind in the east and west of the upper canopy (ZIM-temperature 168 probe, ZIM-relative humidity probe, ZIM Plant Technology GmbH, Hennigsdorf, Germany). On 169 January 25, two additional temperature sensors were installed to cover the northern and the southern 170 sections of the canopy. 171 172 Dry leaf replica experiments 173 174 Eq. 1 was derived under the assumption that the probe is applied to samples containing incompressible 175 water. Furthermore, it is assumed that the elastic coefficients of the leaves are temperature-independent 176 and that the cellular osmotic pressure is constant over a wide turgor pressure range (e.g. Zimmermann 177 et al. 2008). The assumption of constant cellular osmotic pressure holds more or less over a large turgor 178 pressure range because of the relatively high volumetric elastic modulus of the cell walls. Close to the 179 plasmolytic point a significant increase of the cellular osmotic pressure is, however, expected (by 10 – 180 20%). At turgor pressure Pc = 0, Eq. 1 loses validity. Porous samples such as dried plant material, but 181 also filters will show a marked response to ambient temperature changes because of the air trapped 182 within the pores of the sample material. Air has a very low specific heat capacity. For distinction of 183 these effects from real turgor pressure effects measurements were performed on dry Banksia leaves and 184 on different filter and paper types (e.g. 0.2 µm Millipore). 185 186 Transpiration rate and stomatal conductance 187 188 On 7 February, 2012, diurnal measurements of transpiration rate (E) and stomatal conductance (gs) to 189 water vapour were performed at roughly 2 hourly intervals using a portable photosynthesis system (LI-190 6400XT, LI-COR Biosciences, Lincoln, NE, USA). All measurements of gs were conducted at ambient 191 temperature, ambient vapour pressure deficit (VPD), ambient light intensity (PPFD) and 385 ppm leaf 192

5

chamber CO2 concentration. Pre-dawn recordings were taken before 6 am (Australian Western Standard 193 Time, AWST) with the leaf chamber illumination turned off. For subsequent measurements the LED 194 light unit was set to track ambient light levels. Individual recordings were taken as soon as gs had reached 195 steady-state which commonly occurred in less than 3 minutes. 196 197 Pressure chamber measurements 198 199 The diurnal measurement campaign also included measurements of balancing pressure, Pb, using a 200 Scholander-type pressure bomb (PMS Instruments Co., Model 1000, Albany, OR, USA). Sample leaves 201 were harvested in close vicinity to leaves equipped with a ZIM-probe, using a sharp razor blade. Samples 202 were immediately sealed in a zip-lock bag containing a moist tissue and placed into the pressure chamber 203 within 2 min after excision. During measurements the chamber pressure was increased at a rate of ca. 204 0.3 MPa min-1. 205 206 Sap flow measurements 207 208 Sap flow was recorded from 31 January, 2012 until the end of February 2012. We used a heat ratio 209 method (HRM) sap flow sensor (Burgess et al. 2001; ICT International, Armidale, NSW, Australia) 210 installed at 1.3 m height on the south-facing side of the trunk. The HRM sensor was protected against 211 environmental effects using silver foil bubble wrap insulation. Readings were logged every 10 min. The 212 sap velocity data were corrected for the zero offset at night by using the values from the night with the 213 lowest VPD with no observed rainfall. 214 215 Statistical analysis 216 217 All statistical analyses were performed with the free software R, version 2.13.2 (R Development Core 218 Team, 2011; www.r-project.org). We used generalised additive mixed models (GAMM, package mgcv) 219 to investigate the diurnal patterns of leaf physiological parameters and the response of normalised Pp to 220 VPD. Model selection was based on the Akaike Information Criterion (AIC) considering differences 221 between AIC values (AIC) 10 statistically significant (Burnham and Anderson 2002). 222

For the analysis of the diurnal leaf physiological recordings we applied GAMMs comprising the 223 entire day. First we formulated a GAMM with one common smoother term for all cardinal directions or 224 canopy positions (upper or lower), respectively. This model comprised ‘cardinal direction’ and ‘vertical 225 canopy position’ as nominal explanatory variables. To test for differences between cardinal directions, 226 this general model was compared to nested models using individual smoother terms for each cardinal 227 direction or any grouping combination of cardinal directions. To test for differences related to vertical 228 canopy position, the general model was compared to a model with two individual smoothers for upper 229 and lower canopy. 230

For the analysis of the VPD response we binned VPD in 0.1 kPa intervals and derived the 95th 231 percentiles of the Pp values or normalised sap flow within these intervals for the hot and rainless period 232 from 6-13 Feb 2012. To test for differences in Pp-VPD response between cardinal directions and canopy 233 height we compared several models with a smoother term for VPD and ‘cardinal direction’ and ‘vertical 234 canopy position’ (upper or lower canopy) as nominal explanatory variables. Testing for differences 235 between cardinal directions and between upper and lower canopy was based on model comparisons as 236 described above. A mixed model approach was required to include variance structures (varIdent, varExp 237 or a combination thereof, package nlme) in the models to deal with variance heterogeneity. 238

The response of normalised sap flow rates to VPD was modelled by generalised nonlinear least 239 squares regression using a Mitscherlich-type function: 240

241

(Eq. 5) 242

243

m a xe1m a x

SF

VPDcVPD

S FS F

6

where SF is actual sap flow rate, SFmax is maximum sap flow, is the initial slope, VPD is vapour 244 pressure deficit, and VPDc is the VPD at which sap flow reaches zero. In this model we applied an 245 exponential variance structure (varExp) to model variance heterogeneity. 246

For all models, the underlying assumptions were assessed using quantile-quantile plots and 247 histograms to test for normally distributed normalised residuals and plots of these residuals vs. fitted 248 values and each explanatory variable to check for homogeneity of variance. 249 250 Results 251 252 Diurnal course of stomatal conductance, patch pressure, balancing pressure and sap flow 253 254 On a cloudless and sunny summer day in the beginning of February 2012, we performed diurnal 255 measurements of gas-exchange and balancing pressure (Pb) to complement the continuous Pp and sap 256 flow recordings. 257

Across the tree crown, leaf transpiration (E) was more tightly linked to light intensity than to 258 vapour pressure deficit (VPD; Fig. 1). Transpiration rates showed pronounced differences between 259 cardinal directions but the diurnal courses seen in the upper and lower canopy did not warrant separate 260 fits (∆AIC = 24 in favour of the GAMM with individual smoothers for cardinal direction; ∆AIC = 0.7 261 for height-varying canopy models). In east- and north-facing leaves E increased linearly with increasing 262 light intensity and reached peak values of around 6 mmol m-2 s-1 before noon coinciding with maximum 263 light intensity. Peak values of E in the West and South reached maxima at similar times but were about 264 a third lower compared to East and North and showed a more gradual decline (apart from the upper 265 south-facing foliage). 266

The diurnal recordings of gs also showed similar patterns between upper and lower canopy 267 foliage (∆AIC = 21 in favour of the GAMM with one common smoother for vertical canopy position). 268 When testing for differences in gs across cardinal directions in the B. menziesii canopy, the lowest AIC 269 was associated with the model where ‘N+E’ and ‘S+W’ groups were formed (∆AIC = 106.8). Overall, 270 pre-dawn gs was low and did not exceed values of ca. 60 mmol m-2 s-1, irrespective of vertical crown 271 position and cardinal direction (Fig. 1). Maximum gs occurred in the morning between 8 and 10 am 272 coinciding with VPDs of 1.5-2 kPa (Fig. 1). Peak values of gs ranged between 175-250 mmol m-2 s-1, 273 except for west-facing leaves in the lower crown whose gs only reached about 100 mmol m-2 s-1. On 274 average, maximum stomatal conductance across the canopy was 184 ± 10 mmol m-2 s-1 (mean ± SE, n 275 = 25 leaves). For the remainder of the day, gs continuously declined in all leaves and returned to pre-276 dawn levels between 5 and 6 pm. 277

The normalised Pp curves derived from ZIM-probe recordings on upper and lower crown leaves 278 showed the typical diurnal pattern, minimum values during the night (highest turgor), increasing after 279 sunrise (decreasing turgor pressure due to transpirational water loss), peaking between 12 and 2 pm 280 when VPD reached very high maximum values between 5 and 6 kPa (lowest turgor) and decreasing in 281 the afternoon (turgor recovery phase). The Pp curves roughly matched the pattern derived from spot 282 measurements of Pb using a Scholander-type pressure chamber. At all times, Pb values remained below 283 2.5 MPa regardless of vertical crown position or cardinal direction. Differences in Pb values between 284 upper and lower crown largely reflected differences in gs and in the west- and south-facing leaves also 285 the more markedly differing light intensities between upper and lower crown. However, when 286 comparing the entire daily courses, these differences were statistically not significant (vertical crown 287 position: ∆AIC = 16.8; cardinal direction: ∆AIC = 11 in both cases in favour of the ‘one common 288 smoother’ GAMM). While the continuous ZIM-probe measurements allowed tracing the dynamics of 289 leaf water status in response to small changes in microclimate during the day, this variation was certainly 290 not detectable in Pb. Plotting the normalised Pp versus the pooled and averaged Pb data yielded a strong 291 positive linear correlation (r2 = 0.61, P < 0.0001; Fig. 2). Normalised sap flow increased steeply after 292 sunrise and levelled off into a plateau with near maximal values at around 10 am (Fig. 3). The plateau 293 extended to 6 pm and subsequently normalised sap flow declined rapidly to values around 0.45 until 294 sunset. Afterwards, the return to nighttime values occurred much more slowly and was not yet completed 295 at midnight (Fig. 3). 296 297 Pp and sap flow response to VPD 298

7

299 During mid-summer, the VPD response of normalised Pp differed significantly between cardinal 300 directions (∆AIC = 176.9, in favour of the GAMM with individual smoother terms for each cardinal 301 direction; Fig. 4). Pooling cardinal directions into ‘N+E’ and/or ‘S+W’ groups did not improve the 302 model. In the east- and north-facing leaves, normalised Pp increased steeply to values around 1 with 303 rising VPD up to VPD values between 2-3 kPa and then either increased more slowly, remained constant 304 or declined to values around 0.6 (Fig. 4 A–D). West- and south-facing leaves showed a more uniform 305 response, their Pp increased in a more linear fashion with increasing VPD (Fig. 4 E–H). In the south and 306 west, upper and lower crown leaves showed similar responses to VPD (AIC < 1). By contrast, in the 307 east and north the VPD response differed markedly between upper and lower crown due to the decrease 308 of Pp at higher VPD seen in an east-facing leaf from the lower crown and a north-facing leaf in the upper 309 crown (east: AIC = 27.4, north: AIC = 46.1; Fig. 4 B, C). Normalised sap flow showed a typical 310 saturation response to VPD with fairly steady plateau values up to very high VPDs of ca. 8 kPa (R2 = 311 0.97, P < 0.0001). At 2.5 kPa VPD normalised sap flow reached 95% of its maximum (Fig. 4 I). 312 313 Pp leaf gradients 314 315 B. menziesii has highly sclerophyllous, undulate leaves with oblong blades that are 2-3 cm wide and 15-316 20 cm long. In adaptation to the seasonally dry climate and/or nutrient-limitation a thick cuticle has 317 evolved and the stomata are aggregated and sunken in crypts on the abaxial leaf surface (see Fig. S2 318 available as supplementary material to this paper). Simultaneous patch pressure (Pp) recordings at the 319 base, mid and apex of individual leaves gave evidence for the temporary development of Pp gradients 320 within the leaf during day- and/or the nighttime (Fig. 5). The initial clamp pressures of the three ZIM-321 probes on a north-facing leaf were coincidentally almost identical (Fig. 5 B), whereas the initial clamp 322 pressures of the probes on east-, west- and south-facing leaves differed markedly from each other due 323 to the inhomogeneous properties of the leaves (Figs. 5 A, C and D). All ZIM-probe measurements in 324 Fig. 5 were normalized even though the initial clamp pressures of the three ZIM-probes in Fig. 5 B did 325 not necessary require normalisation. The daily courses clearly showed that water loss (i.e. increase in 326 Pp) was sometimes larger at the base (Fig. 5 A), sometimes at the mid (Fig. 5 B) or sometimes at the 327 leaf apex (Figs. 5 C and D). Evaluation of measurements performed over six weeks confirmed this 328 finding. Similar fluctuations in water loss from base to apex were found when the probe located in the 329 mid of the leaf was omitted or clamped on the opposite side of the mid vein thus excluding the possibility 330 of interference of the mid probe with base-to-apex water transport. 331 332 Diurnal Pp measurements 333 334 Typical diurnal Pp measurements were positively correlated with air temperature and inversely 335 correlated with relative humidity (RH; Fig. 6). Consequently, minimal Pp values (highest turgor) 336 occurred during nights and maxima (lowest turgor) around noon (Fig. 6). Data shown are derived from 337 a probe clamped on an east-facing leaf in the lower crown at the end of summer in March 2012. Probes 338 clamped on leaves in other crown locations yielded qualitatively similar results. Data from the last week 339 of March 2012 were selected because the diurnal temperature variations were very similar from day to 340 day. Maximum temperature values at noon ranged between 31°C and 35°C and minimum temperature 341 at night varied between 14°C and 17°C. By contrast, RH values showed a less consistent pattern. The 342 maximum nighttime RH values increased from around 60% on 22 March reaching values of nearly 343 100% on 26 March. The changes in temperature and RH were only partly reflected in the Pp curves. 344 Despite similar daily peak temperatures the corresponding Pp peak values increased continuously and 345 reached maximum plateau values on 26 March. In the middle of the night of 26 March, at constant air 346 temperature of 15 °C and RH > 90%, Pp started to rise dramatically from typical night time levels around 347 5 kPa to 17 kPa at dawn (see arrow in Fig. 6). Such signal increases at night occur when the absolute 348 turgor pressure drops below a certain threshold (< 50 kPa; see Ehrenberger et al. 2012b) resulting from 349 substantial drought stress. In this state, the intercellular air spaces increasingly affect the magnitude of 350 Pp (state II according to the nomenclature of Ehrenberger et al. 2012b). Interestingly, on the following 351 days the increase of Pp disappeared during the night indicating sufficient water supply to the leaves for 352 rehydration. On 8 April, a similar increase in nighttime Pp values at constant air temperature was 353

8

observed (data not shown). Throughout the entire measuring period, state II occurred from time to time 354 (see Table 1), but its frequency increased appreciably in February and April. Temporary drought stress 355 was very often limited to individual branches. For example, during April probes clamped on east- and 356 west-facing leaves in the lower canopy recorded state II more frequently than probes at other canopy 357 locations (Table 1). On a few days, nearly all probes recorded strong drought stress signals indicating 358 that the entire tree suffered from water shortage (e.g. on 1 February, 10 February, 22 February, 26 March, 359 28 April). Overall, state II was observed 177 times between January and April. Interestingly, the 360 occurrence of state II coincided with high RH during the night (80 – 100%). 361

Commonly, drought effects are also manifested by an increase in the length of the turgor 362 pressure recovery phase during the afternoon. However, this was not observed in B. menziesii leaves. 363 Estimates of relaxation times derived from the roughly exponential Pp decline during the afternoon 364 showed rapid Pp recovery within roughly 60 min with little day-to-day variation throughout the 365 measuring period. 366 367 Collapsing Pp readings due to tree desiccation 368 369 Towards the end of October 2011, peak air temperature sharply increased from 28 to 36 °C over the 370 course of a few days inducing high VPD > 4 kPa (Fig. 7). At that time probe readings started to crash in 371 the upper north-facing leaves, closely followed by the signal collapse in the upper east-facing leaves. 372 This process continued with the lower east- and north-facing leaves and the signal collapse occurred 373 somewhat delayed in the west- and south-facing foliage. 374

East-facing leaves in the upper crown initially responded with an increase in Pp peak values over 375 3 days (October 26-28), but then Pp peaks declined in a nearly linear fashion up until 2 November when 376 the signal collapsed (see straight line in Fig. 7 A). Subsequently, only noisy traces were recorded. 377 Nighttime Pp values also declined over the last four days before the signal collapse (Fig. 7 A). In west- 378 and south-facing leaves irrespective of crown location, the pronounced temperature rise caused a 379 continuous increase in Pp peak values until 2 November when the Pp signal suddenly broke down (see 380 solid line in Fig. 7 B). Surprisingly, minimal nighttime Pp values remained largely unaffected by 381 increasing temperature, but during the last two nights prior to the signal collapse, Pp only reached 382 minima shortly before dawn indicating severely delayed leaf rehydration (Fig. 7 B). The signal collapse 383 in the evening of 2 November was followed by a “flatline” trace. The signal collapse indicated the death 384 of the tree. Consistent with the probe readings complete leaf desiccation occurred within a few days. 385 Interestingly, when monitoring desiccated leaves continued post mortem, the signal returned a few days 386 after the death of the tree showing pronounced, diurnal “Pp changes” (Fig. 7 A and B; start at the dotted 387 line). Such changes in Pp may occur if the leaf pore spaces, including xylem vessels, become 388 progressively filled with air during the desiccation process. The trapped air in the leaf patch under the 389 probe expands and shrinks according to changes in air temperature and thus exerts a corresponding 390 pressure on the sensor of the probe. Model experiments using porous filter papers instead of desiccated 391 B. menziesii leaves confirmed this explanation (Fig. 8). When the probes were removed from the dry 392 pale leaves the patch previously covered by the probe magnets was also dry but less pale, presumably 393 due to protection from radiation. Also, the spot was not impressed, i.e. the edge of the patch ended 394 homogenously with the surrounding leaf area. The temperature-dependency of the ZIM-probes was 395 checked in a control experiment after the removal of the probes from the desiccated leaves. The ‘empty’ 396 probes (i.e. the two magnets clamped together without sample in between) were left hanging free at their 397 previous measurement sites in the tree crown where they were exposed to large temperature variations. 398 Temperature changes over a range of 20 °C only resulted in a signal change of ca. 1 kPa (inset Fig. 7 399 B). 400 401 Discussion 402 403 We investigated leaf water status across the crowns of two mature Banksia menziesii trees during spring 404 and summer using the recently developed non-invasive ZIM-probe in combination with a range of 405 conventional techniques to assess the spatial and temporal dynamics of water relations. 406 407 Diurnal course of stomatal conductance, patch pressure, balancing pressure and sap flow 408

9

409 The diurnal measurement campaign revealed a distinct hysteresis pattern between gs and E, which seems 410 to reflect the microclimatic drivers. E strongly correlated with light and thus temperature with peak 411 values starting from noon. By contrast, gs peaked already around mid-morning and progressively 412 declined once VPD exceeded 2 kPa (lacking recovery with decreasing VPD) indicating high sensitivity 413 to evapotranspirative demand. This pattern was consistent across the B. menziesii crown (Fig. 1) and 414 matches the ‘apparent’ feed-forward stomata response described by Franks et al. (1997). Eucalyptus 415 globulus growing in a plantation forest in Western Australia also showed this type of feed-forward 416 control of gs during peak summer (Macfarlane et al. 2004). Certain pairs of aspects (N+E vs. W+S) 417 produced similar gs which is most likely attributable to the fact that these aspects share similar light 418 exposure times and levels (Fig. 1). Overall, strong stomatal control over transpiration prevented 419 excessive loss of water and thus limited the drop of leaf turgor across the crown. This finding confirms 420 the lack of groundwater access at our location as B. menziesii has been reported to rarely show stomatal 421 closure at sites where the roots tap into groundwater (Dodd and Bell 1993). The mean maximum gs of 422 184 mmol m-2 s-1 across the Banksia crown is close to the maximum values of around 150 mmol m-2 s-1 423 reported for deep-rooting Banksia woodland species during summer but is nearly 50% lower than the 424 value quoted for B. menziesii under optimal conditions (Veneklaas and Poot 2003; Drake et al. 2013). 425 As a consequence of the strong stomatal regulation, Pb values peaked around 2 MPa matching peak 426 shoot water potentials for this species quoted by Dodd and Bell (1993). This value also coincides with 427 the figure reported to cause a 50% loss of stem hydraulic conductivity measured in B. menziesii at a 428 similar site (Canham et al. 2009) suggesting that fine-tuned stomatal regulation reduces the danger of 429 complete hydraulic failure. This was confirmed by the normalised sap flow values, which were leveling 430 off into a plateau at the time when stomatal conductance started to decrease on the day of the diurnal 431 measurement campaign (Fig. 3). However, B. menziesii has rather large stomata resulting in a relatively 432 slow stomatal response to changes in VPD and hence prolonged exposure times to steep leaf-to-air VPD 433 gradients making this species more susceptible to leaf hydraulic failure than co-occurring Banksia 434 species (Drake et al. 2013). The minor variations seen in the normalised Pp traces of the ZIM probes 435 indicate stomatal oscillations in response to small, transient changes in microclimate, which have been 436 reported previously (Zimmermann et al. 2010; Rüger et al. 2010). Picking up such minor changes 437 demonstrates the high sensitivity of the magnetic leaf turgor probe and highlights its suitability for 438 detailed studies of stomatal responses and functioning. 439

440 Pp and sap flow response to VPD 441

442 The response of normalised Pp to VPD illustrates the within-crown variability in terms of leaf 443

water balance (Fig. 4). In east- and north-facing leaves that are longer exposed to solar radiation and 444 hence more prone to drought stress than leaves from the other cardinal directions, a levelling off or 445 decline in signal could be observed that coincided with the VPD range over which we assessed down-446 regulation of stomatal conductance during our diurnal measurements (2-3 kPa). However, we have no 447 straightforward explanation for the lack of this stomatal feedback on Pp signals in west- and south-facing 448 leaves. Remarkably, in most of the west- and south-facing leaves Pp values were not at their lowest 449 values at night when VPD was lowest (Fig. 4), which may point to water reallocation within the crown 450 at night from the less sun-exposed west and south to the east and north where insolation is higher and 451 water therefore in greater demand during the day (cf. Westhoff et al. 2009b). We found that sap flow 452 reached 95% of its maximum rate at 2.5 kPa VPD, which again agrees very well with the VPD values 453 marking the onset of stomatal closure, observed during the diurnal measurements. 454 455 Pp leaf gradients 456 457

On the leaf-level, we suspected turgor gradients along the sclerophyllous lamina due to its large 458 dimensions and structural heterogeneity. The installation of three ZIM-probes evenly spaced along the 459 length of the leaf blade confirmed the presence of leaf internal turgor pressure gradients in this species 460 (Fig. 5). However, the occurrence of the highest Pp (i.e. lowest turgor) was not tied to a particular section 461 of the leaf, which may indicate patchy stomatal conductance (Mott and Buckley 2000) or could result 462 from the undulations that may cause different local transpiration rates for similar stomatal conductances 463

10

due to differences in light exposure. Fast-developing turgor pressure gradients within transpiring leaves 464 have recently been reported for the large-leaved species Tetrastigma voinierianum and Musa acuminata 465 (Zimmermann et al. 2008, 2010). Our findings confirm the results of Turner et al. (1984) who measured 466 differences in water potential within large Helianthus annuus leaves using in situ psychrometers. 467 Interestingly, their study also showed that pressure chamber measurements only reflected the water 468 potential of cells near the midrib and did not yield an average for entire H. annuus leaves. 469

470 Diurnal Pp measurements 471

472 The high-resolution ZIM-probe recordings revealed pronounced fluctuations of Pp values at 473

noon and early afternoon which are most likely the result of passing clouds intercepting sunlight and 474 wind gusts transiently removing the boundary layer of the leaves and thus stimulating transpiration. The 475 observed increase in the daily signal amplitude in response to very similar daily temperature patterns 476 (Fig. 6) suggests increasing loss of leaf turgor and has previously been observed in a number of crop 477 species (e.g. grapevine, citrus, olive, banana, canola, wheat cf. Rüger et al. 2010; Bramley et al. 2013) 478 following the termination of irrigation and is thus a clear indication for increasing water shortage. In the 479 cited studies drought stress also caused an upward shift in the baseline signal, i.e. an increase in the 480 minimum values at night, indicating incomplete leaf rehydration. However, B. menziesii rarely showed 481 this phenomenon implying complete turgor recovery overnight. The tight stomatal regulation observed 482 in B. menziesii may often help prevent excessive leaf dehydration and thus facilitate turgor recovery. 483 Given the frequently high relative humidity and concomitant dew formation on leaf surfaces at night, 484 foliar moisture uptake supported by hydrogels (mucilage) is very likely to occur as has previously been 485 shown for a number of tree species from various habitats (Zimmermann et al. 2007). In a cleverly 486 designed experiment using severed forked branches, Yates and Huntley (1995) have conclusively 487 demonstrated the occurrence of horizontal water transport in tree crowns associated with foliar water 488 uptake. Water absorbed by wetted leaves from one arm of a fork was transferred through the branch 489 xylem and progressively improved the water status in the leaves of the second arm, which was kept dry. 490

491 Collapsing Pp readings due to tree desiccation 492

493 Coincidentally, we obtained data from a B. menziesii tree that succumbed to sudden death in 494

early spring (Fig. 7). The monitored tree had been growing on a crest where groundwater levels are well 495 below 30 m throughout the year (Miller and Easton, unpublished data), which is far beyond the reach of 496 B. menziesii roots (Groom 2004). However, given the ample precipitation during October, especially 497 during the last week prior to the tree’s death (54.8 mm rainfall, see Fig. S1 available as supplementary 498 material to this paper), it seems rather unlikely that the studied tree was limited by low soil water 499 availability when maximum daily spring temperatures unexpectedly rose above 35 °C. Therefore, it is 500 more plausible that the sudden increase in temperature and VPD have caused xylem embolisms which 501 in turn have eventually reduced hydraulic conductance to the extent of complete hydraulic failure. This 502 implies higher xylem vulnerability to embolism due to rapid temperature changes in spring compared 503 to summer when such temperature patterns occur more frequently. In the east-facing leaf in the upper 504 canopy of this tree, the decline in Pp peak values during the days prior to its death suggests gradual 505 desiccation (Fig. 7 A). However, with progressive water loss and more air entering the leaf, the system 506 became increasingly compressible rendering equation 1 invalid, i.e. under these conditions, the declining 507 peak signal must not be mistaken as improving leaf turgor. By contrast, the south-facing leaf from the 508 upper crown, seemed better supplied with water: despite increased drought stress at daytime (increasing 509 Pp peak values), complete turgor recovery was achieved at night and only became impaired during the 510 last two nights before the tree’s death (Fig. 7 B). Here, the abrupt signal collapse is probably indicative 511 of sudden hydraulic failure. When all leaf water had been replaced by air in the desiccating leaves, the 512 Pp signal picked up again now only reflecting the temperature-driven expansion and retraction of air in 513 the pore spaces (Figs. 7 and 8). As drought stress signals were first observed in the northern and eastern 514 section of the crown, north- and east-facing foliage should be given particular attention for monitoring 515 and quantifying drought stress in seasonally dry ecosystems in the Southern hemisphere. Sudden, non-516 fire related Banksia deaths without any prior symptoms of deterioration in plant health have been noticed 517 since the 1960s but have not yet been linked to a particular cause (Crosti et al. 2007). Groom et al. 518

11

(2000) also reported sudden temperature-related deaths of B. menziesii but those were associated with 519 unusually high summer temperatures at a site that had been subjected to ground water abstraction. In 520 their case the Banksia trees were already facing summer drought as well as depleted soil water resources 521 due to drawdown but apparently they had been able cope with these harsh conditions until temperatures 522 surpassed a certain threshold for some days. In the light of these and our findings, it appears that 523 temperature and/or VPD seem to play a critical role in pushing B. menziesii beyond their tipping point. 524 With the future prospect of more frequent and longer lasting heat waves, increasing mortality rates in 525 Banksia trees are likely to occur and suggest that, in the long run, B. menziesii will give way to more 526 resilient species shifting not only community structure but – given their dominance in Banksia 527 woodlands – the whole ecosystem into a new state. 528

The mechanisms driving water transport in plants and the links with carbon metabolism and 529 drought are a matter of ongoing debate (e.g. Meinzer et al. 1999a; McDowell 2011). Many trees have 530 been shown to perform multidirectional hydraulic redistribution of soil water resources via their root 531 system, thus enhancing plant water balance and nutrient availability in the upper soil horizon. Our data 532 indicate that horizontal water redistribution may also happen in the canopy making moisture available 533 from crown locations less prone to heat/drought damage to meet higher transpiration demands and 534 mitigate the risk of hydraulic disruption in more exposed crown areas (cf. Yates and Huntley 1995). 535 Future investigations will focus on sectorial watering and shading approaches to elucidate within-crown 536 water (re)allocation in this species. 537

From a conservation perspective, our data imply that the ZIM-probe is a sensitive tool for 538 detecting incipient drought stress in highly sclerophyllous species and is therefore an ideal instrument 539 for evaluating adaptive management strategies to sustain woodland health and natural water resources. 540

541 Acknowledgements 542 543 We are grateful to Botanic Gardens & Parks Authority for granting permission to conduct this research 544 at Kings Park native bushland. Further, we thank Steve Easton for logistical support and David Symons 545 for field assistance during the diurnal measuring campaign. This study was funded by the State Centre 546 of Excellence for Climate Change, Woodland & Forest Health, Botanic Gardens & Parks Authority and 547 a grant of the Europäischer Fonds für regionale Entwicklung (EFRE 80145650; State Brandenburg) to 548 ZIM Plant Technology GmbH. None of the authors has a conflict of interest to declare. 549 550 551 References 552 553 Bramley H, Ehrenberger W, Zimmermann U, Palta JA, Rüger S and Siddique KHS (2013) Non-invasive 554

pressure probes magnetically clamped to leaves to monitor the water status of wheat. Plant and Soil 555 369:257-268. 556

Burgess SSM, Adams MA, Turner NC, Beverly CR, Ong CK, Khan AA and Bleby TM (2001) An improved 557 heat-pulse method to measure low and reverse rates of sap flow in woody plants. Tree Physiology 558 21:589-598. 559

Burnham KP and Anderson DR (2002) Model Selection and Multimodel Inference: a practical information-560 theoretic approach. Springer, New York. 561

Canham CA, Froend RH and Stock WD (2009) Water stress vulnerability of four Banksia species in 562 contrasting ecohydrological habitats on the Gnangara Mound, Western Australia. Plant, Cell and 563 Environment 32:64-72. 564

Canny MJ (1993) The transpiration stream in the leaf apoplast: water and solutes. Philosophical Transactions 565 of the Royal Society of London Series B: Biological Sciences 314:87-100. 566

Canny MJ (1995) A new theory for the ascent of sap—cohesion supported by tissue pressure. Annals of 567 Botany 75:343-357. 568

Canny MJ (2001) Embolisms and refilling in the maize leaf lamina, and the role of the protoxylem lacuna. 569 American Journal of Botany 88:47-51. 570

Clearwater MJ and Clark CJ (2003) In vivo magnetic resonance imaging of xylem vessel contents in woody 571 lianas. Plant, Cell & Environment 26:1205-1214. 572

Connor DJ, Legge NJ and Turner NC (1977) Water relations of mountain ash (Eucalyptus regnans F. Muell.) 573

12

forests. Australian Journal of Plant Physiology 4:753-762. 574 Crosti R, Dixon KW, Ladd PG and Yates CJ (2007) Changes in the structure and species dominance in 575

vegetation over 60 years in an urban bushland remnant. Pacific Conservation Biology 13. 576 Dodd J and Bell DT (1993) Water relations of the canopy species in a Banksia woodland, Swan Coastal Plain, 577

Western Australia. Australian Journal of Ecology 18:281-293. 578 Drake P, Froend R and Franks PJ (2013) Smaller, faster stomata: scaling of stomatal size, rate of response, and 579

stomatal conductance. 64:495-505. 580 Ehrenberger W, Rüger S, Fitzke R, Vollenweider P, Günthardt-Goerg M, Kuster T, Zimmermann U and Arend 581

M (2012) Concomitant dendrometer and leaf patch pressure probe measurements reveal the effect of 582 microclimate and soil moisture on diurnal stem water and leaf turgor variations in young oak trees. 583 Functional Plant Biology 39. 584

Ehrenberger W, Rüger S, Rodríguez-Domínguez CM, Díaz-Espejo A, Fernández J, Moreno J, Zimmermann, 585 D. S, V.L., and Zimmermann U (2012) Leaf patch clamp pressure probe measurements on olive 586 leaves in a nearly turgorless state. Plant Biology 14:1-9. 587

Fernández JE, Rodriguez-Dominguez CM, Perez-Martin A, Zimmermann U, Rüger S, Martín-Palomo MJ, 588 Torres-Ruiz JM, Cuevas MV, Sann C, Ehrenberger W and Diaz-Espejo A (2011) Online-monitoring 589 of tree water stress in a hedgerow olive orchard using the leaf patch clamp pressure probe. 590 Agricultural Water Management 100:25-35. 591

Franks PJ, Cowan IR and Farqhar GD (1997) The apparent feedforward response of stomata to air vapour 592 pressure deficit: information reveaied by different experimental procedures with two rainforest trees. 593 Plant, Cell & Environment 20:142-145. 594

Goldstein G, Andrade J, Meinzer F, Holbrook N, Cavelier J, Jackson P and Celis A (1998) Stem water storage 595 and diurnal patterns of water use in tropical forest canopy trees. . Plant, Cell and Environment 21:397-596 406. 597

Groom PK (2002) Seedling water stress response of two sandplain Banksia species differing in ability to 598 tolerate drought. Journal of Mediterranean Ecology 3:3-9. 599

Groom PK (2004) Rooting depth and plant water relations explain species distribution patterns within a 600 sandplain landscape. Functional Plant Biology 31:423-428. 601

Groom PK (2011) How do trees and shrubs in Perth's Banksia woodlands survive summer drought?, in Perth's 602 Banksia Woodlands – Precious and Under Threat, Urban Bushland Council (WA) Inc., Mount 603 Claremont. p^pp 19-26. 604

Groom PK, Froend RH and Mattiske EM (2000) Impact of groundwater abstraction on a Banksia woodland, 605 Swan Coastal Plain,Western Australia. Ecological Restoration & Management 2:117-124. 606

Heathcote DG, Etherington JR and Woodward FI (1979) An instrument for non-destructive measurement of 607 pressure potential (turgor) of leaf cells. Journal of Experimental Botany 30:811-816. 608

Hinckley TM, Lassoie JP and Running SW (1978) Temporal and spatial variations in the water status of forest 609 trees. Forest Science 24:Suppl.: 1-72. 610

Huber B and Schmidt E (1936) Weitere thermo-elektrische Untersuchungen über den Transpirationsstrom der 611 Bäume. Tharandter Forstliches Jahrbuch 87. 612

Hughes JD, Petrone KC and Silberstein RP (2012) Drought, groundwater storage and stream flow decline in 613 southwestern Australia. Geophysical Research Letters 39. 614

Jones HG (1992) Plants and microclimate. Cambridge University Press, Cambridge. 615 Keighery G (2011) Banksia Woodlands: A Perth Icon, in Perth's Banksia Woodlands – Precious and Under 616

Threat, Urban Bushland Council (WA) Inc., Mount Claremont. p^pp 3-10. 617 Klausmeyer KR and Shaw RM (2009) Climate Change, Habitat Loss, Protected Areas and the Climate 618

Adaptation Potential of Species in Mediterranean Ecosystems Worldwide. Plos One 4:1-9. 619 Lambers H, Chapin FS and Pons TL (2008) Plant Physiological Ecology. Springer, Berlin, Heidelberg, New 620

York. 621 Macfarlane C, White DA and Adams MA (2004) The apparent feed-forward response to vapour pressure 622

deficit of stomata in droughted, field-grown Eucalyptus globulus Labill. Plant, Cell & Environment 623 27:1268-1280. 624

McArthur WM (2004) Reference soils of south-western Australia, (Johnston DAW and Snell LJ eds), 625 Department of Agriculture, Western Australia, Perth. p^pp 285. 626

McDowell N (2011) Mechanisms Linking Drought, Hydraulics, Carbon Metabolism, and Vegetation 627 Mortality. Plant Physiology 155:1051-1059. 628

13

Murphy R and Ortega JKE (1995) A new pressure probe method to determine the average volumetric elastic 629 modulus of cells in plant tissues. Plant Physiology 107: 995-1005. 630

Meinzer F, Clearwater MJ and Goldstein G (1999) Water transport in trees: current perspectives, new insights 631 and some controversies. Environmental and Experimental Botany 45:239-262. 632

Meinzer F, Goldstein G, Franco AC, Bustamante M, Igler E, Jackson P, Caldas L and Rundel PW (1999) 633 Atmospheric and hydraulic limitations on transpiration in Braziliancerrado woody species. Functional 634 Ecology 13:273-282. 635

Milburn JA (1996) Sap ascent in vascular plants: challengers to the cohesion theory ignore the significance of 636 immature xylem and the recycling of Münch water. Annals of Botany 78:399-407. 637

Mott KA and Buckley TN (2000) Patchy stomatal conductance: emergent collective behaviour of stomata. 638 Trends in Plant Science 5:258-262. 639

Passioura JB (1991) An impasse in plant water relations? Botanica Acta 104:405-411. 640 Rüger S, Netzer Y, Westhoff M, Zimmermann D, Reuss R, Ovadya S, Gessner P, Zimmermann G, Schwartz 641

A and Zimmermann U (2010) Remote monitoring of leaf turgor pressure of grapevines subjected to 642 different irrigation treatments using the leaf patch clamp pressure probe. Australian Journal of Grape 643 and Wine Research 16:405-412. 644

Sommer B and Froend R (2011) Resilience of phreatophytic vegetation to groundwater drawdown: is recovery 645 possible under a drying climate? Ecohydrology 4:67-82. 646

Tobiessen P, Rundel PW and Stecker RE (1971) Water potential gradient in a tall Sequoiadendron. Plant 647 Physiology 48:303-304. 648

Turner NC and Long MJ (1980) Errors arising from rapid water loss in the measurement of leaf water potential 649 by the pressure chamber technique. Australian Journal of Plant Physiology 7:527-537. 650

Turner NC and Sobrado (1983) Evaluation of a non-destructive method for measuring turgor pressure in 651 Helianthus. Journal of Experimental Botany 34:1562-1568. 652

Turner NC, Spurway RA and Schulze ED (1984) Comparison of water potentials measured by in situ 653 psychrometry and pressure chamber in morphologically different species. Plant Physiology 74:316-654 319. 655

Veneklaas EJ and Poot P (2003) Seasonal patterns in water use and leaf turnover of different plant functional 656 types in a species-rich woodland, south-western Australia. Plant and Soil 257:295-304. 657

Westhoff M, Reuss R, Zimmermann D, Netzer Y, Gessner A, Geßner P, Zimmermann G, Wegner LH, 658 Bamberg E, Schwartz A and Zimmermann U (2009a) A non-invasive probe for online-monitoring of 659 turgor pressure changes under field conditions. Plant Biology 11:701-712. 660

Westhoff M, Zimmermann D, Schneider H, Wegner LH, Gessner P, Jakob P, Bamberg E, Shirley S, Bentrup 661 F-W and Zimmermann U (2009b) Evidence for discontinuous water columns in the xylem conduit of 662 tall birch trees. Plant Biology 11:307-327. 663

Yates DJ and Hutley LB (1995) Foliar uptake of water by wet leaaves Sloanea woollsii, an Australian 664 subtropical rainforest tree. Australian Journal of Botany 43:157-167. 665

Zimmermann D, Reuss R, Westhoff M, Geßner P, Bauer W, Bamberg E, Bentrup F-W and Zimmermann U 666 (2008) A novel, non-invasive, online-monitoring, versatile and easy plant-based probe for measuring 667 leaf water status. Journal of Experimental Botany 59:3157-3167. 668

Zimmermann D, Westhoff M, Zimmermann G, Geßner P, Gessner A, Wegner LH, Rokitta M, Ache P, 669 Schneider H, Vásquez JA, Kruck W, Shirley S, Jakob P, Hedrich R, Bentrup F-W, Bamberg E and 670 Zimmermann U (2007) Foliar water supply of tall trees: evidence for mucilage-facilitated moisture 671 uptake from the atmosphere and the impact on pressure bomb measurements. Protoplasma 232:11-34. 672

Zimmermann U, Rüger S, Shapira O, Westhoff M, Wegner LH, Reuss R, Geßner P, Zimmermann G, Israeli Y, 673 Zhou A, Schwartz A, Bamberg E and Zimmermann D (2010) Effects of environmental parameters and 674 irrigation on the turgor pressure of banana plants measured using the non-invasive, online monitoring 675 leaf patch clamp pressure probe. Plant Biology 12:424-436. 676

677 678

Figure legends 679

680 Fig. 1 Diel courses of transpiration rate (E), stomatal conductance (gs), normalised patch pressure (Pp), 681 given as Pp*/Pp, max*, balancing pressure (Pb, using a Scholander-type pressure chamber), photosynthetic 682

14

photon flux density (PPFD), and vapour pressure deficit (VPD) in the upper (dotted lines, open circles) 683 and lower (solid lines, filled circles) canopy recorded on February 7, 2012. Means ± SE, n = 3-7, apart 684 from patch pressure where n = 1 (one ZIM-probe per cardinal direction and vertical crown position) and 685 pressure bomb measurements where n = 2. Note, that VPD was nearly identical in the upper and lower 686 crown and hence pooled. AWST denotes Australian Western Standard Time. 687 688 Fig. 2 Relationship between normalised Pp values and corresponding balancing pressure values Pb 689 derived from pressure chamber measurements performed on 7 February, 2012. Symbols represent mean 690 values averaged over all cardinal directions and the upper and lower canopy. 691 692 Fig. 3 Diurnal course of normalised sap flow in B. menziesii on 7 February, 2012. AWST denotes 693 Australian Western Standard Time. 694 695 Fig. 4 The response of normalised patch pressure (Pp) in upper and lower leaves from all cardinal 696 directions (A–H) and normalised sap flow (I) to vapour pressure deficit (VPD) during a rainless period 697 in peak summer 2012 (February 6-13). 698 699 Fig. 5 Simultaneous leaf patch clamp pressure (ZIM-probe) recordings along the lamina of B. menziesii 700 leaves in the upper canopy of a ca. 5 m tall B. menziesii tree. ZIM-probes were installed at the base 701 (black line), mid (blue line) and apex (red line) of leaves facing all cardinal directions (A: east, B: north; 702 C: west and D: south). The lower panels in each plot show concomitant ambient air temperature (T, 703 solid lines) and relative humidity (RH, dotted lines). The ZIM-probe output signal is referred to as 704 normalised patch pressure Pp*/Pp,max*. Owing to different initial patch pressures among probes on a 705 leaf, values were normalised as described in the materials and methods section (data in B did not 706 necassary require normalisation because the initial clamping pressure was coincidentally virtually 707 identical for the three probes at the day of installation). For the sake of clarity the diel Pp curves were 708 smoothed using a 9-point FFT filter routine. Grey rectangles indicate nighttime hours. AWST denotes 709 Australian Western Standard Time. 710 711 Fig. 6 Typical diel patch pressure (Pp) measurements together with the corresponding changes in 712 ambient air temperature (T, solid line) and relative humidity (RH, dotted line) measured in close vicinity 713 to leaves equipped with ZIM-probes. Data were extracted from a 4-month recording starting at the end 714 of January 2012. Probe readings were derived from an east-facing leaf in the lower crown. Note the 715 increase in Pp during the night of 26 to 27 March (arrow), which is an indication of temporarily severe 716 drought stress. Grey rectangles indicate nighttime hours. 717 718 Fig. 7 Diel recordings of patch pressure (Pp), air temperature (T) and relative humidity (RH) in the 719 crown of a ca. 6-m tall B. menziesii tree from October 25 to November 9, 2011, comprising the period 720 prior to the tree’s death (solid line) and measurements recorded after the death of the tree. The dotted 721 line marks the point when the leaves were completely desiccated and the diurnal “Pp changes” started 722 again caused by the trapped air in the leaf patch under the probe. Note that probe readings collapsed 723 several days earlier in upper east-facing leaves (A) than in upper south-facing leaves (B). Diel Pp 724 readings collapsed in the following order: Nupper leaves, Eupper leaves, Elower leaves, Nlower leaves and then the lower 725 and upper west- and south-facing leaves simultaneously. Inset: ZIM-probe readings without any sample 726 material between the magnets. 727 728 Fig. 8 ZIM-probe readings together with air temperature (T) and relative humidity (RH) on a 0.2 mm 729 Millipore filter recorded at the same crown location at the beginning of December, 2011. The “Pp” data 730 reflect the expansion and shrinkage of air-filled spaces (vessels etc.) within the desiccated leaf. “Pp” 731 changes measured on other types of filter and typewriting paper were much smaller indicating that the 732 air-filled pore structure plays an important role. Grey rectangles indicate nighttime hours. 733 734 Supplementary Fig. S1 Daily minimal/maximal temperature (Tmin, Tmax) and precipitation for 735 winter/spring 2011 and summer/autumn 2012 from the Perth ‘Metro’ weather station (Australian Bureau 736 of Meteorology, 31.92° S, 115.87° E, 25 m a.s.l.) approximately 5 km from the study site. Grey 737

15

rectangles indicate the experimental periods for the study trees 1 (deceased) and 2. The dotted line 738 indicates the turn of the year. 739 740 Supplementary Fig. S2 Cross section of Banksia menziesii leaf embedded in glycomethacrylate and 741 stained with toluidine blue indicating hair filled stomatal crypt on abaxial surface (A) thickened cuticle 742 on adaxial surface (B) and sclerenchyma cells (stained blue) that account for the rigid structure of the 743 leaf (C). 744 745 Table 1: Frequency of Pp night increases (state II) from 22 December, 2011 to 30 April, 2012 measured 746 on leaves of a ca. 5 m tall B. menziesii tree in each cardinal direction in the upper and lower crown 747 (Note: each row in the table reflects a different branch in the crown). 748 749 750

December January February March April

East

lower leaf

1

2 0 5 4 14

lower leaf

2

- - 6 1 3

upper leaf

1

2 0 5 0 2

upper leaf

2

- - 2 0 3

Nort

h

lower leaf

1

2 0 4 0 3

lower leaf

2

- - 2 5 4

upper leaf

1

2 0 3 0 1

upper leaf

2

- - 3 0 3

Wes

t

lower leaf

1

2 1 5 3 13

lower leaf

2

- - 3 2 7

upper leaf

1

1 0 2 1 7

upper leaf

2

- - 5 2 1

Sou

th

lower leaf

1

1 0 1 0 2

lower leaf

2

- - 2 4 14

upper leaf

1

2 0 4 1 4

upper leaf

2

- - 5 2 13 751

East North West South

●

●

●

●●

●●

●

●

●

●

●

0

2

4

6

E (m

mol

m−2

s−1

)

●

●

●

●

●

●●

●

●

●●

●

●

● ●

●

●

●●

●

●●

●

●

●

●

Upper canopyLower canopy

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

● ●

●

●

●

●

●

●

●

0

50

100

150

200

250

g s (m

mol

m−2

s−1

)

●

●

●

●

●

●

●

●●

●●

●●

●

●

●●

●

●

●●

●

●●

●

●

●

●

●

●

●

●

● ●●

●

0

40

80

120

160

P* pP

* p,

max

(%)

●

●

●

●

●

●

●

●

● ●

●

●

0

0.5

1.0

1.5

2.0

2.5

Pb (M

Pa)

●

● ●●

●

●

●

●

●●

●

●

●

●

●

●

●

●●

●

● ●

●

●●

●

●

●

●

●

●

●●

●

●

●

0500

100015002000

PPFD

(µm

ol m

−2 s

−1)

0

2

4

6

06 10 14 18

VPD

(kPa

)

06 10 14 18 06 10 14 18

Time (h) (AWST)06 10 14 18

●●

●

●●

●

●●

●

●●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

●

●●

●

●

●

●

●

●

●

●

0

50

100

150

0 0.5 1.0 1.5 2.0 2.5

r2 = 0.61 P < 0.001P

* pP

* p, m

ax (%

)

Pb (MPa)

06 10 14 18 22

0

0.2

0.4

0.6

0.8

1.0

Nor

mal

ised

sap

flow

Time (h) (AWST) 7 Feb 2012

Upper canopy Lower canopy

East

�����

��������

��

�

�

�

���

������

���������

�

���

�

����

���

�������������������������

���

0

0.2

0.4

0.6

0.8

1.0A

� Leaf 1Leaf 2 �

�

���������

�������

��������

�

�

���������

���

�

�

�

���

��

��

�

���

��

�

���

���

�

����

��������

��

���������

�������

��������

�

�

���������

���

�

�

�

���

��

��

�

���

��

�

���

���

�

����

��������

B

North

������������

���

�������

����

���������������

������

��������������������

0

0.2

0.4

0.6

0.8

1.0

Nor

mal

ised

Pp

C

���

���������������

���

�����

���������

����

�

����������

�

��

�

�

������������

���

���������������

���

�����

���������

����

�

����������

�

��

�

�

������������

D

West

�������

��

����������������������������

������������

���

��������������

0

0.2

0.4

0.6

0.8

1.0E

��

�����������������������

���

������

�����

�������

�����������

���������

F

South

���

�

��������

����

�������

��

�

��

�

�

�

�

�

��������

�

����������������

���

��

0

0.2

0.4

0.6

0.8

1.0

0 2 4 6 8

G

���

��������

������������������

���������

�

�

�

�

����������

�����������

0 2 4 6 8

H

0

0.2

0.4

0.6

0.8

1.0

0 2 4 6 8

���

�

�

��

���

�

�����

��

����������

��������

����

��������

��������������

����� �

���

R2 = 0.97

P < 0.0001

Nor

mal

ised

sap

flow

VPD (kPa)

I

2.5

Winter Spring Summer

0

10

20

30

40

50Te

mpe

ratu

re (°

C) Tmax

Tmin

0102030405060

J J A S O N D J F M A

Prec

ipita

tion

(mm

)

Month (2011−2012)

Tree 1 Tree 2